Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:01
The Art of Leadership Network. We've
0:03
already seen the impact that COVID and
0:05
the smartphone has had on young people's
0:07
ability to form lasting social bonds
0:09
with one another, to have
0:12
deep conversations and deep relationships. And
0:15
who knows already what knock on impact that's going
0:17
to have 10 years down the road to the
0:19
population, let alone anything else. If young men can't
0:21
find young women to build the
0:23
population with without some other artificial intervention,
0:25
we're going to have problems, guys. That's
0:27
the way it's going.
0:29
And to the ability for you to have
0:32
a virtual girlfriend or a virtual boyfriend and
0:34
fall in love with the chatbot and be
0:36
totally happy with that, I think
0:38
is a relational category that we don't want
0:40
to see emerge. And that's actually probably the
0:42
more likely worst case scenario that
0:44
we are already seeing companies
0:47
building tools, building experiences where you
0:49
can essentially take a step back
0:51
from society and have pretty much
0:54
everything done for you, both
0:56
your work product, your relational product, your
0:59
call it whatever aspect of your life
1:01
you want managed. And we
1:04
end up in a kind of, you remember that
1:06
Disney movie, Wally, right? We kind of end up
1:08
in that kind of all sat slightly in a
1:10
half vegetative state looking like a potato leaning back,
1:12
watching a recycling truck clean up the world. That's
1:16
not a great outcome, generally
1:18
speaking. Welcome
1:23
to the Kerry Newhoff Leadership Podcast. It's
1:26
Kerry here and I hope our time
1:28
together today helps you thrive in life
1:30
and leadership. Man, I am excited to
1:32
talk about all things AI with James
1:35
JP Polter. He has started
1:37
and sold an AI company. We talk
1:39
about, well, everything for the church and
1:42
he's involved in that space. And
1:44
believe it or not, if you've been like a late
1:47
adopter, hey, we got some really practical tools to get
1:49
you started. If you like me are interested in the
1:51
meta issues as well, like what is this going to
1:53
do to us? We can talk about
1:55
that and a whole lot more. Today's episode is brought
1:57
to you by my live event, The Art of Life.
2:00
of Leadership Live. I'd love for you to
2:02
join me in Dallas, Texas September 16th to
2:04
18th. So if you're
2:06
ready to break through what's
2:08
been holding you back,
2:10
you can secure your
2:12
spot by visiting theartofleadershiplive.com.
2:14
That's theartofleadershiplive.com. And today's
2:16
episode is brought to you by my friends
2:19
at 10 by 10. You know they are
2:21
committed to making faith matter more to the
2:23
next generation. You can visit 10X10. That's t-e-n-x-1-0.org
2:28
slash R-D-I to complete a
2:30
free assessment that will measure your
2:32
youth ministry's efforts. So that's 10
2:35
by 10.org/R-D-I to complete
2:37
your free assessment today. Well,
2:40
what is the best case scenario for
2:42
AI in the next decade? What's the
2:44
worst case scenario? What are some practical
2:46
tools you could start using now? And
2:49
what are the pastoral implications of AI?
2:51
I sit down with James Polter. JP
2:54
is the head of AI and innovation
2:56
at House 337, former
2:59
CEO of Vixen Labs, which sold to House 337
3:01
in December of 2023. Vixen Labs is one of
3:06
the leading consultancies focused on
3:08
conversational AI. JP is
3:10
also the founder of Eclizia, a think
3:12
tank focused on the future of AI
3:14
for the church. He was previously the
3:17
head of emerging platforms and partnerships at
3:19
the Lego Group, cool job, where he
3:21
set up the likes of
3:23
Lego Life, the company's social network
3:25
for children, and oversaw a
3:27
number of the group's partnerships with Meta,
3:30
Spotify, and many more. He's
3:32
a sought after international speaker,
3:34
podcaster, and writer on the
3:36
future of AI and voice
3:38
assistance and innovation culture. So
3:40
really delighted to have him on the podcast
3:42
today. Hey, leaders, if you're looking to level
3:45
up your leadership and level up your church,
3:47
I'm hosting my very first conference, The Art
3:49
of Leadership Live in Dallas,
3:51
Texas from September 16th
3:53
to 18th. So conference is going to be
3:56
a little bit unconventional instead of listening to
3:58
eight hours of keynotes each day. day
4:00
and leaving with pages of notes that never
4:02
turn into real results. You've been there, right?
4:04
You get done those conferences. The
4:06
Art of Leadership Live has a really
4:08
cool balance of teaching, connection, and free
4:10
time so you can find the right
4:12
insights with the right people and act
4:14
on them. So yeah, I'm going to
4:16
be giving some talks, but we're
4:19
going to have open and honest discussion with me and
4:21
with other people who are there. Practical
4:23
takeaways. It's an intimate event. We cap
4:25
registration at a very low number. So
4:28
to ensure the right people are there
4:30
with you, the event is by application only.
4:32
There are only a limited number of spots.
4:35
It is close to being full. So
4:38
act now. You can go
4:40
to theartofleadershiplive.com to learn more
4:42
and register before it's sold
4:44
out. Go to theartofleadershiplive.com. Register
4:47
now before it's too late. And then
4:49
in my conversations with a lot of
4:51
you, you've shared about the challenges you're
4:53
facing with engaging young people in your
4:55
church. I get it, man. It's hard.
4:57
I know you're trying everything from outreach
4:59
activities to small group, but do
5:01
you know whether you're actually making a difference
5:04
or not? Well, our friends at 10 by
5:06
10 are committed to making faith matter more
5:08
to the next generation. They're
5:10
offering you a free five-minute
5:12
assessment called the Relational Discipleship
5:15
Inventory. And after you complete the
5:17
survey, you'll get an instant assessment
5:20
that measures your youth ministry's efforts
5:22
against the seven discipleship emphases that
5:25
are proven to help you grow
5:27
in relational discipleship, radically focused on
5:29
Jesus and his love for the
5:32
next generation. So you can
5:34
visit 10by10.org/RDI. That's
5:38
t-e-n-x-1-0.org/RDI. To
5:41
complete your free assessment today,
5:43
that's 10by10.org/RDI, and you can
5:45
get your free assessment. And
5:47
now my conversation with JP
5:50
Polter. JP, welcome to the
5:52
podcast. Okay, thanks so much for having me. It's a
5:54
pleasure to be here. Yeah, so I wanna start in
5:56
the deep end. What
5:58
are the threats? of AI.
6:00
There's like two different views, right? There's
6:02
the benevolent view of AI, and then
6:05
there's the malevolent view of AI. So
6:07
from where you sit, what are the
6:09
threats, the existential threats that AI poses
6:11
right now? Well, I
6:13
think the biggest distinction is that I'm not
6:15
sure that AI is the existential threat, but
6:17
it's AI in the hands of humans that
6:19
might be the existential threat, which
6:22
is probably an important distinction. In
6:24
the church, we have a habit, sometimes a nasty
6:26
one when it comes to technology of looking for
6:29
666 in the code of everything that
6:31
we use, assuming that there's
6:33
something evil lying behind it. I
6:36
don't think that that's what's happening here, but we
6:38
obviously know from all the work that's being done
6:41
across the industry that AI does have the potential
6:43
to radically transform society and particularly when
6:45
put in the hands of those that may not
6:47
want to use it for good, can
6:49
really have some pretty devastating
6:51
impacts on things like the economy, on
6:54
politics, and on the geopolitical
6:56
space. There's a real opportunity that
6:58
things could go awry with AI
7:01
being used by bad actors. But AI itself
7:03
doesn't seem to, at the moment, want to
7:05
come and kill us all, even
7:07
though it might have the potential to do so. So
7:10
I don't think that we are starting the
7:12
conversation from a position that we should be
7:15
fearful of AI. But we are
7:17
in a position where we need to be
7:19
faithful with it and particularly take a considered
7:21
approach to how we use it as Christians and how we
7:23
bring it into the church and everything else that we do.
7:26
One of the things that I
7:28
think about a lot is unintended
7:30
consequences. So if you look at
7:33
social media a decade ago, most people,
7:35
well, maybe not a decade ago, but
7:38
15 years ago as it was developing,
7:40
most people would say, oh, this is
7:42
good. We had no idea it would
7:44
produce the unintended consequences, particularly among teenage
7:46
girls. Gen
7:48
Z with the anxiety, Jonathan Haidt
7:50
has done some incredible work in
7:52
that area this year. Just highlighting.
7:55
I've just got finished reading The Anxious
7:57
Generation. I think we see that those...
8:00
like second and third order consequences just
8:02
couldn't have been anticipated when that technology
8:04
emerged. Well, exactly. And I mean, you
8:06
know, I didn't see it either
8:09
and I was an avid adapter and you
8:11
know, now I realize, oh yeah, this is
8:13
messing with my brain too. So,
8:15
you know, when you think about AI in the
8:18
future, one of these sub
8:20
arguments under how social media is used but
8:22
also AI is it's all
8:24
monetized. Like there's government policy can hardly
8:26
keep up. The EU has done probably
8:28
the best job globally or worst job
8:30
depending on how you look at it
8:33
of regulating technology. But like the
8:36
government doesn't even understand it. And
8:38
I was reading some backstory on Sam
8:40
Altman being kicked out of the
8:43
board for OpenAI earlier this
8:45
year and then being
8:47
brought back in after he was offered
8:49
by Microsoft and it seemed to be
8:52
the backstory there from what I
8:54
can determine. And again, I don't know Sam Altman
8:56
and I don't live in Silicon Valley is
8:59
that it was his seeming lack
9:01
of regard for the
9:04
human consequences of AI and really
9:06
the race to be first to
9:08
market, dominant market, profit, profit, profit.
9:11
Any thoughts on that when really you
9:13
can make an argument with the magnificent
9:15
seven that it's just
9:17
a race to shareholders, greatest profitability to
9:20
be first, et cetera. Is that what
9:22
you mean by the existential threat being
9:24
in the hands of humans? Or
9:26
what's your take on monetization and all
9:29
of this? Well, so
9:31
the existential threat problem mostly comes from
9:34
these models getting out of hand and
9:36
into the hands of those that might
9:38
use them for nefarious purposes. We think
9:40
about things like fake news, mass generation
9:43
of content that may affect
9:45
elections. Obviously we're in the year, as
9:47
we record of the elections. And
9:49
depending on when this comes out, we'll see how many of
9:52
those it's been affected by AI. But I would guess a
9:54
lot. Whether or not there's a
9:56
real effect or at least a correlatory
9:58
effect that we point out. and
10:00
say, hey, there was fake news in the
10:02
ecosystem. It'll be hard to make an argument
10:04
with anybody that that's not really been the
10:06
case. Whether or not it's had any real causal
10:08
effects on the outcome of
10:10
elections, that's yet to be seen. But certainly, that it'll
10:13
be out there and it'll be something that people point
10:15
at. And that in itself is disruptive, right? When
10:18
something like this new technology comes into the space,
10:20
long before, think back to Cambridge Analytica, think
10:22
back to what happened with Facebook. Long before anyone found
10:25
out there was something going
10:27
on under the hood there, people were still
10:29
pointing at Facebook and pointing at many other
10:31
social platforms and saying, this is going to
10:33
have a negative effect on our elections or
10:35
on the outcome of other social issues. And
10:38
I think we can already do that with AI
10:40
in an unchecked, unregulated way. Now, the
10:42
good thing is I do think global regulators
10:44
are learning from what's happened in the social
10:46
media space. And as you
10:48
say, in the EU and in the US and around
10:50
the world, people will move fast to
10:52
try and bring some kind of regulation in. But
10:55
the big difference here is that this isn't a
10:58
channel that we're talking about. AI isn't something
11:00
that has just become one person's job in
11:02
the company or one person's job in the
11:04
church. This affects everything. It's affecting all the
11:07
tools that we use. Now, even the things
11:09
we're using to record this podcast or the
11:11
computer that you're using today, certainly
11:13
the phone that you're going to use in the next 24 months is
11:16
going to have some kind of AI living inside of it. And
11:18
until we all have a better understanding of what that
11:21
means for us, we're not going to be turning it
11:23
off by default. We're most of the time going to
11:25
be opted into it. And so we
11:27
could accidentally, if we're not careful, sleepwalk into
11:29
another type of experience that we've had with
11:31
social media where we are falling foul to
11:33
giving our data over to platforms that may
11:35
want to use it for different purposes, even
11:38
if that's just to monetize it with us
11:40
not knowing about it, even if that's just
11:42
the worst thing that happens, that's still something
11:44
that could be happening to us that we
11:46
might not want to be participating in. And
11:48
I think that's why we've got to be
11:50
cautious. Strange
11:53
analogy here, but I
11:55
just interviewed William Urie, who will
11:57
be on the podcast. So he wrote the infamous Getting Out
11:59
of the World. and TS Fisher and Yuri, and
12:02
has been involved in really
12:05
the biggest negotiations in the last 50 years. And
12:07
although we didn't spend a lot of time on
12:09
it in the conversation, remember talking
12:11
to him off mic about the arms race, like
12:13
the start talks, the salt talks. You're probably too
12:15
young to remember those in real time. I was
12:18
in college, I remember them. Right about them. And
12:20
there was the arms race, right? Like in the
12:22
80s, that was a very real thing. And
12:25
it looked like, you know, there was
12:27
the mad pack, the mutually assured destruction that
12:30
if the United States or the Soviet Union,
12:32
now Russia, press a nuclear button,
12:35
the planet was blown to smithereens
12:37
in minutes. And we came close
12:39
a few times and then that
12:41
decelerated. So do you see
12:43
right now, we can have another race
12:46
in technology to the top. It's profit
12:48
driven. It's not driven by states and
12:51
governments, it's driven by private
12:53
industry. Do you see
12:55
any kind of, and I know you work in
12:57
this area, hence the question, any kind of, not
13:02
external ethical counsel, but internal
13:04
where OpenAI sits down with
13:07
Anthropics, sits down with the
13:10
thousands of startups, with Meta and
13:12
says, hey, what are we doing
13:14
here that will benefit humanity?
13:16
Do you see any of that happening? Or
13:18
is that really left to outside actors? Well,
13:22
I think that it's definitely coming because the pressure
13:24
is there for them to all be on the
13:26
same page. The challenge obviously with the commercial side
13:29
of it is that none of them want to
13:31
share their toys, right? They all wanna keep their
13:33
own little special source. And that's understandable too. I
13:36
mean, the AI wars, they're basically becoming a
13:38
proxy for the cloud wars that we've seen
13:40
over the past decade of, the Amazons, the
13:42
Googles, the competing, but the difference is that
13:44
they've all been leapfrog by some of these
13:47
startups that they couldn't have anticipated at the
13:49
time. And hence the investments that are coming
13:51
in left, right and center. What's really
13:53
interesting, if you look at the board makeup of
13:55
many of these companies, they've got people sitting across
13:57
them from all these different companies on
13:59
each other's. boards, they're building up different teams.
14:01
You've got people like from Google that are
14:03
now on the Anthropic board, you've got Amazon
14:05
investing in them as well as Google investing
14:08
in them. And so
14:10
it does make the landscape quite muddled, but
14:12
they all are essentially moving towards the same
14:14
thing of for profit as the means of
14:16
growing these things. And there's nothing necessarily intrinsically
14:19
wrong with that. But
14:21
the accountability is the thing that we want to
14:23
see. And obviously, that's the work that's being done
14:25
by the EU with the AI act that's just
14:28
been passed, is what's going through
14:30
Congress at the moment. I'm sure what will be coming
14:32
out in both the Canadian and the UK governments
14:34
as they try and bring something that's of parity
14:37
to those registrations
14:39
and legislations. But
14:41
the challenge is how will we as a
14:43
society respond to this? I think that's the
14:46
bigger problem because governments, they're
14:48
great at regulating big companies. It's much easier for them
14:50
to do because there's less people they have to talk
14:52
to. But I kind of look at
14:54
it as the analogy of when we wanted to
14:56
make cars safer, we had to do two things
14:58
at the same time. You had to go to
15:00
Ford and GM and Volvo and everyone and say,
15:02
hey, put seatbelts in the car. And
15:04
that was one part of the problem. And then the other part
15:07
of the problem was telling all of us to put the seatbelts
15:09
on when we got in the car. And
15:11
that's the bit that seems to me is
15:13
missing. We're not seeing an awful lot of
15:15
work by governments, regulators, but other public bodies
15:18
to make sure that Kerry knows what he's doing
15:20
with his data and JP knows what he's doing
15:22
with his data and just put that on a
15:24
higher pedestal than we do right now. And
15:27
we see this all across the place. And
15:29
you mentioned Johnson Heights work around
15:31
the anxious generation. We're just handing over
15:33
so much of our data in the
15:35
form of our smartphones. And particularly we're
15:37
handing over those smartphones to our kids
15:40
without training them on, hey, this is what's
15:42
going on with the data that you're giving
15:44
away when you sign that little tick box
15:46
without reading the terms and conditions. That's
15:49
where we need to be paying more
15:51
attention. So does anybody read the terms
15:53
and conditions? Like seriously, I know, I
15:55
know. And then it's accept
15:58
all cookies, decline all cookies. or
16:00
worse, manage your preferences. And I've hit the
16:02
manage your preferences. And I'm
16:05
led into this confusing landscape that
16:07
I'm just like, all right, I'll
16:09
just accept them. What's at stake
16:11
in moments like that where we're
16:13
accepting terms and conditions and
16:16
cookies. And I know that Google's gonna
16:18
change that massively this year
16:20
as well with the death of third party cookies. Well,
16:23
I mean, the history of the
16:25
internet is us giving up privacy
16:27
for utility, right? If we find
16:29
enough utility in anything, we'll give up some privacy
16:31
for it. Every time that you want an Uber
16:34
to arrive and find you on the street, you're
16:36
giving up a little bit of privacy for that
16:38
utility. The same if you want your Shopify account
16:40
to remember your login details, or if you want
16:42
to use meta products to browse what your college
16:44
roommates are up to. Every time
16:46
we do that, we're giving up some level
16:49
of privacy for that social utility back. And
16:51
the history of, as I say, the internet
16:53
says that we will continue to do that
16:55
because as humans, just by our very nature,
16:58
we're quite lazy animals. We like to find straight
17:00
paths between point A and point B. And
17:03
as long as we're those lazy animals,
17:05
we'll continue to find ways, even self-inflicted,
17:08
of exploiting that laziness to
17:10
get what we want. And that's where
17:12
the biggest problem is, is that most often
17:14
these products are not particularly clear about what
17:16
they're doing with the data. And
17:18
even if they are clear with what they're doing with
17:20
the data, they're moving at such a speed that we
17:22
can't really know five, 10 years from now what
17:26
this is gonna look like. The AI that we
17:28
have today, as our mutual friend,
17:30
Kelly Zhang says, is the dumbest it will
17:32
ever be. Yes. Right? Yeah.
17:35
And that's the thing that we can't anticipate. If you even think
17:37
about the advancements we've seen in the past 48 months from
17:41
these quite simple language models, things like
17:43
Alexa, things like Google Assistant, then we
17:45
get the first version to chat GPT
17:47
that can just about write something interesting.
17:50
But then these leaps and bounds to the video
17:52
and the imagery that they're able to create. If
17:55
that curve continues to grow exponentially, five
17:57
years from now, 10 years from now.
18:00
now, we can't even imagine what these
18:02
things are going to be able to do. And
18:04
they're all learning from the data that we're giving
18:07
them right now. And they'll
18:09
learn from the data that we create using
18:11
these tools in the years to come. And
18:14
so that's why we need to take more
18:16
accountability for what we're using them for. Well,
18:18
this is relevant on an individual user basis,
18:20
but also for all the leaders who run
18:23
organizations whose organizations are collecting this kind of
18:25
information too, right? So two
18:28
questions for you. Number one, what do you do?
18:30
Do you accept all cookies? Do you decline some?
18:32
Do you manage your preferences? What do
18:35
you do personally with that? Knowing what you know,
18:37
which is probably more than
18:39
most of us listening to this podcast and
18:41
certainly more than the person conducting this interview.
18:45
I think you're more educated than you let on.
18:47
Oh, I do. I would say. A little knowledge
18:49
is dangerous, J.G. It's very dangerous. I
18:52
try. I think I try, but I'm not going
18:54
to say that I'm not guilty of this myself
18:57
because I too want all of those things. But
18:59
I would be particularly when it comes to these AI
19:01
tools that are now being built on top of the
19:03
language models, let's move like the social networks and other
19:06
things like the iTunes privacy agreement
19:08
to one side and think
19:10
about particularly these AI tools. One
19:12
of the things I'm always doing is looking at what
19:14
models are they built on top
19:16
of. You've obviously got things like chat
19:18
GPT, which we all know many people
19:20
might have tried out something like Claude
19:23
or perplexity. And then there's the new
19:25
AIs that tend to feel more friendly
19:27
like the Pi from inflection. Some
19:29
of these tools you guys may find or you'll find in the
19:31
show notes, I'm sure. But these new
19:34
types of tools, they're not
19:36
always the experience that you're having
19:38
with them at least is not always built
19:40
by that same company. We're seeing lots of
19:42
startups, lots of people building new experiences that
19:45
you can use on your phone or on
19:47
the desktop that are wrappers on top of
19:49
other language. It's like a skin, right? Exactly.
19:52
It's a skin with some prompt engineering
19:54
and... Specialized utility.
19:57
If you don't know what model is underneath that...
20:00
you don't know where your data is ultimately going to
20:02
and what it's being used for in the future. Now,
20:04
personally, I don't think there's anything particularly
20:07
nefarious going on at any of these
20:09
companies right now. There's no
20:11
evidence to suggest that from what I'm seeing.
20:13
They all ultimately, as you've alluded to before,
20:15
are commercial enterprises and they want to make
20:18
money. The minute that they lose
20:20
our trust is the minute they stop making
20:22
money. There
20:24
is that kind of interesting tension that the
20:26
commercial aspect of all this holds. It
20:29
doesn't mean that the companies building on top of these things don't
20:32
want to do things that either are not, like
20:35
I say, bad in themselves. But they may
20:37
not be things that we necessarily want to
20:39
support as a society and particularly within
20:41
leadership or within church contexts. These
20:44
are the issues that I think we want to pay more
20:46
attention to because just because something can be done
20:49
doesn't necessarily mean that it should be done or
20:51
certainly that we would want it to be done.
20:54
So when you think about the future, I like
20:57
to think in terms of trajectory
20:59
or trendlines, JP. So
21:01
if you look at the decline of the church
21:03
over time, there's a trendline. And
21:05
even after COVID, I was reading a few graphs
21:07
recently. Basically there's an interruption pattern
21:09
interruption, and then it continues on the
21:12
historic trendline down, which is really too
21:14
bad. Other companies like the
21:16
rise of tech, market capitalization or user
21:18
share. The trendline is here's 2010, here's 2015,
21:20
here's 2020, here's 2024. So
21:26
with no current, no future intervention
21:28
beyond what we see now, I'd
21:32
like to go into the worst case for
21:34
AI and the best case for AI. And
21:37
I know, ask 10 people, get 10 opinions. I get
21:39
it. But I would love your take as
21:41
somebody who's literally built and sold companies in
21:43
AI and who sits
21:45
on a board and has your
21:47
resume. What is the worst case
21:50
that we could imagine a world
21:52
five, 10 years down the road in terms of
21:54
AI and where it's going? Well,
21:57
if anyone has watched the recent Mission Impossible.
22:00
movies, they'll have seen what looks
22:02
like potentially the worst-case scenario, which
22:04
is some kind
22:06
of large-scale AI emerges as
22:08
being... I would
22:11
not say sentient because I'm not sure
22:13
that that's actually possible within our theology,
22:16
but something that emerges with a super
22:18
intelligence that is beyond our control and
22:21
starts deciding that either it's going to
22:23
align itself with those that
22:26
would seek to see the downfall
22:28
of society or the end of the Western world or however
22:30
you want to put it, or that
22:32
it just makes that decision in and of itself
22:34
and says, you know what, humans, you're not doing
22:36
a great job looking after this planet. Maybe we
22:38
would be better off without you. And
22:40
that argument has been made by many. The
22:44
likelihood of that happening, I think, is far
22:46
off because I think there are many steps
22:48
between here and there before we get anywhere
22:50
near that being a challenge. But it's not
22:53
entirely impossible. And
22:55
because it's not entirely impossible, we should at least give it
22:57
some level of attention, if
23:00
nothing else, to make sure it shapes the way that
23:02
we follow policy and that we keep away from those
23:04
things in the same way that we did with nuclear
23:06
weapons, which is often the comparison.
23:10
And probably more importantly, the way that we didn't
23:12
really do it with social media, which is probably
23:14
a more accurate comparison in terms of the actual
23:16
net impact that it's had on people around the
23:18
world. So I think in terms of worst case
23:20
scenarios, that's kind of out there in the existential
23:22
threat category. But what I think is more the
23:25
likely worst case scenario is
23:27
that these models continue to grow
23:29
unchecked and that we just begin
23:31
to adopt them into society in a way that we
23:33
just wouldn't want to see happen. I think the one
23:36
area that I'm particularly passionate about is in
23:38
relationships. We've already seen the
23:40
impact that COVID and the smartphone has
23:42
had on young people's ability to form
23:45
lasting social bonds with one another, to have
23:48
deep conversations and deep relationships. And
23:51
who knows already what knock on impact that's going to
23:53
have 10 years down the road to the population, let
23:56
alone anything else. If young men can't find
23:58
young women to build population
24:00
with without some other artificial intervention, we're going
24:02
to have problems, guys. That's the way it's
24:04
going. Add to
24:06
the ability for you to have a virtual
24:09
girlfriend or a virtual boyfriend and fall in
24:11
love with the chatbot and be totally happy
24:13
with that. I think it's a
24:15
relational category that we don't want to see
24:17
emerge. And that's actually probably the more likely
24:19
worst case scenario that we are
24:21
already seeing companies building
24:24
tools, building experiences where you can
24:27
essentially take a step back from society
24:29
and have pretty much everything done for
24:31
you, both your work product,
24:33
your relational product, your call it whatever
24:37
aspect of your life you want
24:39
managed. And we end up
24:41
in a kind of, you remember that Disney movie,
24:43
Wally, right? We kind of end up in that
24:45
kind of all sat slightly in a half vegetative
24:47
state looking like a potato leaning back, watching a
24:50
recycling truck clean up the world. That's not a
24:53
great outcome, generally speaking.
24:56
And that's before they all had like AI friends,
24:58
they at least were talking to one another. So
25:00
I think that's where we could
25:03
end up in some form. And that
25:05
is to me a more likely worst
25:07
case scenario that necessarily something
25:10
releases all the nukes or turns off all of
25:12
the energy grid. Yeah. And so sort
25:14
of a general social
25:16
degradation over what we have now.
25:19
And for regular listeners, Scott Galloway
25:21
is on this here again. I
25:24
don't know exactly when compared when all the
25:26
episodes will be released, but he sees that
25:28
as a massive threat
25:30
to our culture is disengaged
25:33
young men. He says that's
25:35
where things go bad.
25:37
All right, now be the optimist. I'm
25:39
sure you've got an optimistic framework. Again,
25:41
all things being equal, nothing in the
25:44
future radically changing from the current trajectory.
25:46
What is the possible best case scenario?
25:49
Well, I think this is interesting. You recently
25:51
have Morgan on your show, I think talking
25:53
about kind of the things that don't change.
25:55
Yes. Yes, I'm going to have him. Yeah.
25:58
Or you're going to have him on and yeah. And
26:00
you know, he is
26:02
his recent book, I think is really fascinating because
26:05
he told about the things that that don't change
26:07
over time. And I think that's
26:09
actually more likely. There's a bunch of human characteristics
26:11
that don't change as much as we're worried about
26:13
things like a girlfriend's and stuff. I actually think
26:15
that the overwhelming majority of us still find
26:18
the pull to be with one another in
26:20
the physical space. The real
26:22
optimistic view of AI is that it can
26:24
just take away so many of the drudgery
26:27
tasks that we all have to deal with.
26:29
Yeah, particularly in the ministry
26:31
context or work context. You
26:34
know, we spend so much time administering
26:36
the work we do rather than doing
26:38
ministry. Or we spend so much time
26:40
on the busy work that is often
26:42
called rather than actually being busy working
26:45
on the thing that we want to focus on. And
26:47
AI does hold the promise to be able
26:50
to eliminate vast amounts of that work, days,
26:52
if not weeks, a year of that, you
26:54
know, filling in a spreadsheet or entering
26:56
a form for the 17th time that was built
26:59
on Windows 97 or whatever it might be. Like
27:01
these things are out there still.
27:04
And as much as the technical world has
27:06
been revolutionized, you walk into any public municipal
27:08
building or hospital or school and you'll find
27:10
technologies that have been leapfrogged time and time
27:13
again still running in, you know, many of
27:15
these systems. And AI
27:18
potentially could come along and if not replace those
27:20
technologies, just fill those things in for us, which
27:22
would make things a lot easier for most of
27:25
our lives. So administration, I
27:27
think, is going to be one of the big things that
27:29
maybe we do end up in the world of being able
27:31
to do the four day work week and, you know, living
27:33
that kind of dream that Tim Ferriss has been selling to
27:35
us for a number of years. That we
27:37
actually might have some lifestyle, you know,
27:40
benefit over time. I think that's one of the big things. But
27:43
it's also imperative, right? Like if we want to
27:45
solve some of the biggest issues in the
27:47
world, climate change, cancer, other
27:49
kind of diseases, and
27:52
also make sure that our politics is actually something
27:54
that can sustain itself over time. We
27:56
haven't done a great job of that as of late as the
27:58
human race since the industrial era. revolution. And
28:01
those problems are now so big that the
28:03
speeds required to fix them doesn't seem to
28:06
be possible in and of ourselves. And
28:08
so my optimistic view would be that we
28:10
will see massive breakthroughs whether it's the work
28:12
that's being done on things like protein folding,
28:15
or the discovery of new drugs, the
28:17
identification of new cancers, the solving of
28:19
the climate race, or emergence of
28:21
things like small nuclear
28:24
fusion and nuclear fission reactors, which will give
28:26
us boundless energy sources. We're not going to
28:28
get to those things about artificial intelligence.
28:30
We don't have the collective brainpower on
28:32
our own. And so I
28:34
think that we could begin to see the
28:36
new renaissance begin to emerge over the next
28:39
decade as these technologies get
28:41
smarter, so long as the negative
28:44
opportunities don't outweigh the positive ones,
28:47
and that people continue to seek
28:49
for human flourishing rather than the
28:52
rather commercial, sad
28:55
story that is being told at the
28:57
moment, just flooding the internet with garbage
28:59
content. That's what seems more likely right
29:01
now. But I'm hopeful that
29:03
we might see a new generation
29:06
of technologists arrive that say, wow, we can
29:08
use this stuff to do immense good. We'll
29:11
probably toggle between the macro and the micro,
29:13
but to dip down into the micro for
29:15
a moment, what are some of
29:17
the most interesting uses
29:20
of AI that you're seeing right now?
29:22
And I'd love you to take that
29:24
in two phases. Number one, just what's
29:26
on your phone, what's on your laptop
29:29
that you're finding absolutely fascinating on
29:31
a micro level, and then other
29:34
technologies that perhaps you haven't discovered
29:36
personally yet, but you know are out there.
29:38
What's got your attention? Well, I mean, the
29:41
app that probably gets more attention on my
29:43
phone, if I think about it here than
29:45
any other, is perplexity
29:48
at the moment. As a search engine, it
29:50
has almost entirely replaced my use of Google.
29:52
This is not an endorsement for them. I
29:55
don't have a particular affiliation. Nothing
29:58
of that nature, but I would just say. from
30:00
straight out usage, it's amazing
30:02
how quickly that has replaced things.
30:06
A story from this week, a client
30:08
rang us out of the blue at Vixen Labs, the
30:10
agency that I run, saying,
30:12
hey, I got in touch with you because I found
30:14
an article about the very specific thing I was looking
30:16
to do with AI this week. And you were the
30:18
top answer ranked on complexity because it
30:20
was found very specifically the article he
30:23
was looking for in a way that
30:25
Google had no ranking for us whatsoever.
30:28
And even if you take that in
30:30
microcosm, it's a really good example of
30:32
how these new tools are helping find
30:34
much more contextual information. So it's amazing.
30:38
If nothing else, because I can ask it
30:40
questions and ask it follow-up questions and it
30:42
seems to do a better job of looking
30:44
at stuff than I can do on my
30:46
own. Just Google search has so degraded over
30:48
the last couple of years. I know they
30:50
keep making changes, but it's not functionally useless,
30:52
but it's approaching that. And
30:55
once these Gemini models begin to roll out
30:57
into mainstream Google search, which we can only
30:59
anticipate they will, then I think there's a
31:01
real chance Google catches up. But for now,
31:03
for me, I spend an awful lot
31:05
of time in perplexity for anyone that
31:07
makes that slide where or content for the
31:10
internet, which is sure is all of us
31:12
that get stuck in PowerPoint from time to
31:14
time. I've been using tools like beautiful.ai, which
31:16
has an amazing ability for, you know, if
31:19
you're a pastor or an executive leader of
31:21
some description right now having to put together
31:23
this year's annual report or quarterly earnings, being
31:26
able to design a slide deck which
31:28
has the Claude model from Anthropic running
31:30
inside of it. You can
31:32
create an entire 20 slide presentation with one
31:35
line text prompt. It does
31:37
the layouts for you. It goes to Dali and
31:39
creates images for you. It chooses the most appropriate
31:41
slide layout like it's an amazing tool. Again, not
31:43
aligned with them, but just that's the one that
31:46
I spend literally hours in
31:48
every single day. But perplexity and beautiful.ai.
31:51
Yeah, for sure. Those are two. I'll be checking those out. Those
31:53
are new to me. I've heard of them, but I haven't done
31:55
anything with them. Yeah, they're definitely great
31:57
places to start. And if you are already using.
32:00
something like chat GPT, you can access the
32:02
GPT-4 models, you can access the Claude models
32:04
inside of complexity as well. So you can
32:06
switch around and pay one subscription price. Is
32:09
it just me or is Claude significantly
32:11
more intelligent than chat GPT-4 right now?
32:15
It's certainly more eloquent, not necessarily more
32:17
intelligent, which is really interesting. What
32:20
we're beginning to see with some of
32:22
these models, they're bringing personality in some
32:25
ways. Claude model is particularly good at
32:27
long form written content. That's probably why
32:29
I like it. Yeah, well
32:31
for you exactly and for myself as well. It's
32:34
much better at emulating my voice when I
32:36
give it examples than chat GPT is. And
32:39
it sticks to the task for longer
32:41
than kind of wandering off and doing
32:43
something strange, which GPT can kind of
32:45
be prone to do. So yeah, I
32:47
would definitely say Claude is also
32:49
a really good place to play around with, particularly
32:51
if you're a content creator, if you're a writer,
32:53
if you're someone that writes press releases or social
32:55
media content or biographies or whatever
32:58
it might be, anything that's longer form,
33:00
Claude is probably your go-to model right
33:02
now for doing that well. And what's
33:04
really promising is those models are getting
33:06
small enough to be able to run
33:08
on a phone offline, which opens up
33:11
some really interesting opportunities in places where
33:13
connectivity is bad or
33:15
parts of the world where AI tools
33:17
would be pretty hard to justify
33:19
just because of the amount of data that they use. So
33:22
I'm particularly excited about where they might go.
33:25
Yeah, that's interesting. So again, a one-on-one
33:27
question, and by the way, for Canadian
33:29
listeners, most are American. You
33:31
can't get Claude, at least at the time of recording
33:33
this, but a VPN will work you around that pretty
33:35
easily. Yeah, well, yeah. And you can
33:37
also access it via some of these other tools as well,
33:39
which is again, the thing we were
33:41
saying before about some of these tools, they have
33:44
other models inside of them. So yeah, your mileage
33:46
might vary, but play around. And a VPN is
33:48
a great idea anyway on public internet, but beyond
33:50
that, that's a virtual private network. Claude, by the
33:53
way, is C-L-A-U-D-E.AI. And
33:57
what it, okay, so help me
33:59
understand as. somebody who does not code
34:01
and someone who is not into AI deeply. ChatGPT
34:04
is an LLM, Clode is an LLM
34:06
as well, large language model. Is that
34:08
right? But they're trained on
34:10
different things. Like I don't understand, because
34:12
they are very different. Like
34:14
user interface, Clode is like, oh, that
34:16
was really helpful. ChatGPT is one more
34:19
time please with feeling. You know? Yeah.
34:21
So it's helpful to understand the difference
34:23
between a large language model and the
34:25
app that you might use that thing
34:27
in. Yeah. Clode
34:29
has, there are a number of different Clode
34:31
models. It doesn't help that they're called Clode
34:33
as well, but there are a number of
34:35
different Clode models. If you go to clode.ai,
34:37
you can interact with the Clode model or
34:40
at the time of recording, it may be
34:42
a different one from when you're listening to
34:44
this, but the Clode 3 model or Clode
34:46
4, Clode 5, whatever comes down the line.
34:48
So the language model
34:50
itself is trained on a massive
34:52
amount of human created data, some
34:54
synthetic created data living on the
34:56
internet. And depending on who you're
34:59
getting that model from, they've sourced it from
35:01
different places. Some have more commercial agreements in
35:03
place, others are just scraping stuff that's in
35:05
the public domain. But the big
35:08
difference is the way that they're trained and tuned
35:10
to behave in different ways. And then when you
35:12
come and use them inside of a tool, for
35:14
example, inside of ChatGPT, you're using
35:16
what is probably the GPT 3.5 or GPT 4 model. Those
35:21
different models have different amounts of data in
35:23
them, different knowledge, and they behave in slightly
35:25
different ways, but you can access both of
35:27
them inside of the ChatGPT app. And
35:30
it's the same models that if you
35:32
use BingChat, for example, or you're using
35:34
Microsoft Copilot, you're using the GPT 4
35:37
model inside of those tools.
35:40
And they have different abilities too. So Bing Search is-
35:42
That's sort of the stream idea, right? Yeah,
35:44
but so Bing Search uses the same model that
35:46
ChatGPT has, but it has the ability to go
35:48
to the internet and pull in real time results
35:51
and blend it with the content that the
35:53
model can produce. The same thing that Perplexity
35:56
is able to do when they're using their
35:58
own Perplexity search model and clause. isn't
36:00
able to do in and of itself because
36:02
it doesn't have live internet access. So they
36:05
all have slightly different abilities. It's
36:07
kind of like having a range of different assistants that
36:10
you might employ to different jobs. Some
36:12
of them are going to be better at different things. And
36:14
so chat GPT is particularly good at tasks
36:16
and handling data. It's particularly good at writing
36:18
code. Claude is much better
36:21
at linguistic tasks, long form textual
36:23
tasks. And neither of
36:25
those produce images, but that's why you have
36:27
things like DALI from OpenAI or Mistral.
36:30
And there are many others that can produce
36:32
visuals as well. Okay. So just
36:36
from a basic thing, Anthropic, which
36:38
is the company behind Claude, are
36:42
they working totally independently from
36:45
chat GPT? There's no master data source.
36:47
These are two separate companies that have
36:49
developed themes from scratch. As far as
36:51
we're aware. As far as
36:53
we're aware. Let's not get into conspiracy theories.
36:55
Well, no, but it is an important thing
36:58
to highlight that none of these companies publish
37:00
exactly what data has been trained to make
37:02
them work. Not in full
37:04
anyway, even the open source ones. Now you
37:06
can begin to get an understanding of what
37:09
might be in there because we've seen examples.
37:11
And this is the basis of the court
37:13
case being held by the
37:15
New York Times against OpenAI at the
37:17
moment is that clearly chat GPT is
37:19
able to produce, well, at least was
37:22
able to produce long form examples of
37:24
what seemed to be exactly copies of
37:26
New York Times articles. Therefore, you can
37:28
assume it was in the training data.
37:31
It may no longer be. And
37:33
so each of these different models, they have slightly
37:36
different training data that goes inside of them. And
37:38
that's what makes them better or worse at certain
37:40
tasks. There are some models that are trained, for
37:42
example, just on scientific papers. And so scientific papers,
37:44
if you're going to write them, they have a
37:46
very specific linguistic style, a way of referencing, et
37:48
cetera. And so you would want
37:50
a model that was trained on that. And this is where we see
37:52
a lot of companies going right now is
37:55
building on top of these language models with
37:57
their own data sets. So you've seen examples
37:59
from Blue. Bloomberg doing it with financial data,
38:01
or McKinsey, the consulting firm, doing it with
38:03
all of their consulting content. And we're now
38:06
seeing in the Christian world, Bibles being built
38:08
on top of just seeing the release this
38:10
past couple of weeks of Bible.ai, which is
38:12
a project that's trying to build a large
38:14
corpus of Christian content to
38:16
train a foundational language model so it's
38:19
better at producing biblical related content. So
38:21
we're going to see examples of that
38:23
type of thing emerge as well. It'd
38:25
be fun to see him merge so
38:27
many different theologies into one that'll be
38:29
absolutely fascinating to watch. And that's
38:31
the problem, right? Because first of all, you want to
38:33
put something in them, you've got to agree with what
38:35
we agree on. Which is
38:38
not always the easiest of things. Okay, I'm going
38:40
to have to track that one. Now that's a
38:42
good survey of the landscape, it really is, JP.
38:45
You specialize in conversational AI.
38:47
What is that? So
38:50
yeah, so we started Vixen Labs, which is the agency
38:52
I've run for the past five or six years, really
38:55
with intent around making experiences
38:58
of talking with the companies and organizations that
39:00
you have to do business with every day
39:02
a little bit better through
39:04
talking to them rather than having
39:06
to click and scroll and swipe
39:08
down. Is this like instead of
39:10
press five? Precisely. And
39:12
so we've been really focused on things
39:15
like voice technologies, we spend a lot
39:17
of time building applications for Alexa, for
39:19
Google Assistant over the past couple of years. And
39:21
now that's emerged into these large language models
39:24
that can help us build things like chatbots
39:26
and voice assistants that are just much more
39:28
pleasurable to use and actually helpful than
39:31
they have been historically, more than just
39:33
doing timers and reminders, which
39:35
is what they've been good for for a little while. So
39:37
yeah, we focus on that because I think
39:39
we want the future of using the internet
39:41
to be one that's just a little bit
39:43
more personal and more human,
39:46
ironically. Which
39:48
we are conversational animals, right? That's how
39:50
we get things done. You don't
39:52
roll up a McDonald's drive-thru in order
39:56
to have a chat with someone. But
39:58
the means that you do so... the
40:00
means you get that burger delivered into the car,
40:02
or whatever your order is, is through
40:05
the art of conversation. And that's the way
40:07
we've always done things as humans. We believe,
40:10
certainly from a Christian perspective, we
40:12
began with the word, right? And say, we
40:14
know that conversation and the art
40:16
of talking to one another is an important thing. So
40:19
I would love to see more of our technology
40:21
head in that direction, rather than it being this
40:23
kind of screens in front of us at all
40:25
times head down in a phone. We
40:28
think that by bringing the ability to talk to
40:30
and type to if necessary, tech,
40:33
it lifts us out of that trap and
40:36
lifts us out of that doom scrolling mentality. Oh,
40:39
that's a great thing. So for those who've
40:41
called Apple in the last six, seven years,
40:44
you'd be familiar with that kind of conversational
40:46
AI where it's like, I'm a fully trained
40:48
voice specialist that can help you. What's wrong?
40:50
And then you tell it, my computer won't
40:52
boot up. And it's like, oh, it sounds
40:54
like you need technical support. Hang on. I'll
40:56
put you right through that kind of thing.
40:59
Well, so that's the, I think what we would see as
41:01
conversational tech, like 1.0 is from natural
41:05
language processing and being able to
41:07
understand natural language understanding. So
41:10
that is the press one for this. And it
41:12
knows at least what you said. But the difference
41:14
now we see is with these language models is
41:16
that you could say the entire thing of what's
41:18
wrong with your Mac book. And
41:21
if Apple chose or other manufacturers chose to
41:23
put these systems in place, it
41:25
could do what chat GPT could do, which is
41:27
take all of that information from you directly and
41:29
actually give you the answer. It would probably work
41:32
out pretty well because it's the same
41:34
technology. It's the same technology. It's just being spoken to
41:36
you rather than showing you as a line of text
41:38
on screen. And we're seeing this now if you use
41:40
the chat GPT app on your phone, you can talk
41:42
to it and it talks back to you in a
41:44
variety of different voices. And you can always flick away
41:47
and look at the screen. But actually
41:49
that experience of talking to these things with
41:51
your voice is actually quite delightful and
41:54
certainly feels far more human even though
41:56
it's not than typing
41:58
into a screen or into a text box. Yeah.
42:01
All right. I want to talk about because
42:03
you work with Ecclesia,
42:07
Ecclesii. How do you pronounce it? Ecclesii.
42:10
Ecclesii. I think the pronunciation we got
42:12
in was about, you know, sing it written down is different. It's
42:15
a think tank that is starting to
42:17
process or that is processing the theological
42:19
issues associated with AI. What
42:22
is, if you were to look at
42:24
an agenda of what consumes your time
42:26
and Ecclesii's time,
42:28
what are the top two or
42:31
three things on your agenda right now? What
42:33
are you working on? So
42:35
I think, Ecclesii, we founded the think tank at the end
42:37
of 2023 as we began to see the
42:41
emergence of the church really responding
42:43
to the AI revolution and particularly
42:45
for church leaders in all contexts,
42:48
whether that's in ministry, in
42:51
parachurch organizations, in non-profits, but Christian
42:53
leaders looking for guidance on how
42:55
do I wrestle with some of
42:57
these big topics that are arising.
42:59
And broadly speaking, we've rallied around three big
43:02
areas. One is to have a better theology
43:04
of AI because it does raise some existential
43:06
questions. Maybe we can get into those. I know you've done a
43:09
lot on that in the past. But
43:11
then the practical and the pastoral implications of
43:13
the other two real areas. So
43:16
practically, what does it mean to have an AI
43:18
policy for your church or for your charity? What
43:20
does it mean to have a code of ethics? How
43:23
will you physically go around choosing which tools to
43:25
use and which ones that sit outside of things?
43:28
And then pastorally, using that as a kind
43:30
of broad catchall, is how you
43:32
deal with the issues that arise from AI
43:35
in ministry or in society. And I always use
43:37
this example, which I know is maybe a bit
43:39
trite, but when a young person comes to a
43:41
youth worker and says, hey, I've fallen
43:44
in love with this chatbot. What do
43:46
I do? How does a
43:48
youth worker or a minister in that context
43:50
respond to that question? What do
43:52
they actually say? Do they acknowledge the feelings which
43:54
may be very real for that young person who
43:56
feels like they truly have a relationship with these
43:59
things? In recent
44:01
months of people sadly committing suicide when
44:03
these bots break up with them or
44:05
that they have tried to marry them,
44:07
that we've seen both the good
44:09
and the bad of these things and that's already now, let
44:12
alone 24 months from now when this stuff gets really smart.
44:15
So how do we contend with these issues? This
44:17
is the stuff we're trying to wrestle with because
44:20
even if you choose not to, quote
44:22
unquote, use AI in your ministry, your
44:25
congregations, those that you're trying to seek to
44:27
reach and be impacted and certainly on the
44:29
mission field, they will be using it. There's
44:31
no way that we're putting this back in
44:33
the box. And so what we really want
44:36
to see at Ecclesiast is the church be
44:38
prepared and equipped to respond to that situation
44:40
and to be able to roll with these things
44:43
as they emerge, not be caught on the back
44:45
foot as we've so often seen it be the
44:47
way in the past when it comes to technology
44:49
and the church where we're often bad at it
44:51
or often late or often both. And
44:54
I think we could do a better job of it this time around.
44:57
I really appreciate it. And that example of
44:59
somebody falling in love with an AI or
45:01
a chatbot has been used before on this
45:03
podcast and it sounds
45:05
like science fiction, but it's not
45:07
science fiction. Yeah. So I
45:10
mean, one of the things that we see
45:12
and I saw this at a hackathon that
45:14
I was at recently where there was an
45:16
example of a church organization that was trying
45:18
to create a technology using CCTV and
45:20
AI and it could do face tracking to monitor
45:22
attendance. Now there's probably a bunch of good reasons
45:24
why you might want to do that, but
45:27
there's also a lot of challenges with it as well,
45:29
right? Totally doable with
45:31
existing technology like plug together a few
45:33
different tools. Totally doable right now. Should
45:36
it be done? I think real
45:38
problematic use case because
45:41
the minute you start tracking who's there, who's
45:43
not, who's to stop someone calling them up
45:45
and saying, Hey, your giving looks like it's
45:47
dropped this past month and you've not been
45:49
at church three times or worse. Hey,
45:52
I saw you sit next to the same woman three
45:54
times in the past month in church and she's not
45:56
your wife and your wife wasn't there. Is there something
45:58
going on there? Like these are... Again,
46:01
it's not about the AI being
46:03
nefarious. It's AI in the hands
46:06
of someone that might use it in the wrong way that is
46:08
the problem. These are real-time
46:10
things and we need the church to be not
46:12
just going, oh, isn't this exciting? I
46:14
can do things that, well, I don't have to have someone
46:17
at the back with a clicker registering how many people showed
46:19
up this Sunday. I can just have an AI do it.
46:21
Well, where is that data going? What are you
46:23
going to do with that data? What are the
46:26
ways you're going to keep that person safe, let
46:28
alone the safeguarding implications and many other issues besides?
46:31
These types of things are already happening.
46:33
The technology is already there to use
46:36
it. I want us to use
46:38
this technology. I don't want to sound like I'm saying put the
46:40
brakes on, but you wouldn't get in
46:42
a car with a great accelerator with no brake pedal.
46:45
We need to have both. That's what we're advocating
46:47
for is we need to go
46:49
forward and try and find use cases for this stuff,
46:51
but also know when we need to apply the brakes
46:54
and make sure that we're driving on the right road. Well,
46:57
and I think that's a good point. I
46:59
mean, that product probably isn't at market right
47:01
now and hopefully not being developed, but
47:04
it probably will be and it probably will be
47:06
released by somebody down the road, whether it's out
47:08
of that source or a different source. And
47:10
then you need the wisdom to figure it out.
47:12
I was thinking when you were saying the unintended
47:15
and the intended consequences of AI are in your
47:17
church. Imagine a church in 2008, just
47:22
as Twitter was emerging, Facebook was starting to
47:24
escape out of campuses saying, we're not doing
47:26
social media. We're not doing, as we called
47:28
it then, Web 2.0. It's banned
47:30
in our church. We won't have an account if
47:33
that happened. And let's say you're still reaching people
47:35
in 2024, which is a whole other question. Everybody
47:39
in your church, you're dealing with
47:41
increased teen anxiety, depression, polarization,
47:44
extremism, all of those things have
47:46
happened to your church, even
47:49
though you're off social media. And
47:51
your argument is you can, and a surprising number,
47:53
I mean, Glue has done the studies on that
47:56
Glue and Varna, surprising number of people in
47:58
the church are opposed to AI. AI and
48:00
saying, we're not going to use it. Well, first of all,
48:02
you're probably already using it and don't realize it. Secondly, even
48:05
if you did that, that doesn't mean your church isn't
48:07
going to have a whole bunch of people, like meaningful
48:09
percentages in love with chatbots,
48:12
using technology for malevolent purposes, not knowing
48:14
ethically what to do as a Christian
48:16
about how to integrate it in their
48:18
businesses and in their home, what to
48:21
do with their teenagers, their marriage broke
48:23
up, um, over AI. Like
48:26
all that stuff is coming in here. It's all
48:28
coming. And it's all coming out. It's also coming,
48:30
you know, for the church leader that might be
48:32
listening, it's also coming into the actual delivery of
48:34
ministry as well. Right. Yeah. We, we get this
48:36
question all the time about, can I use chat
48:38
GBT to help me write a sermon? And
48:41
I know that we can because you, you and
48:43
many other works on building helpful tools to help
48:45
you do exactly that because it can't be done.
48:48
Um, but you know, should we be
48:51
asking it to write all of our prayers for a Sunday
48:54
and just reading them the base and you know,
48:56
should we be asking it to compose
48:58
worship music and using that without the
49:00
spirit being part of that process? Like
49:02
these are really thorny theological issues that
49:04
we have to try and unpick. And
49:07
I don't think most churches let
49:09
alone have time to think about this stuff
49:12
or the, or the knowledge or education to
49:14
do so. And nor should they, they're not
49:16
AI theologians sitting around thinking about this stuff,
49:18
we need to equip them with the, with
49:20
this knowledge. Um, but yeah,
49:22
it's very much starting at the baseline of do
49:24
people know what the tools are? Do
49:26
they know how they're being used? And
49:28
crucially, are we thinking critically about them
49:30
before we actually pick up and log
49:32
in before you click yes to the
49:34
terms and conditions again, without reading them.
49:37
Um, just because the thing looks like it's
49:39
going to save you two hours of an
49:41
afternoon, like these are the steps we need
49:43
people to take and to be aware that,
49:46
um, when we add technology to
49:48
anything, we give it the opportunity
49:50
for it to take over some of our decision-making,
49:52
which we might want, but we also give it the
49:54
opportunity to take over some of our mistakes, which we
49:56
don't want. And that's the thing that
49:59
we want to be conscious. of is that if you
50:01
wouldn't be willing to give this task to an assistant
50:04
without checking what they were doing, don't give it to
50:06
an AI to do it either. We
50:08
need to stay in the loop and help use
50:11
these tools responsibly, particularly in the ministry context
50:13
if we want trust in the church to
50:15
increase, to use your analogy
50:17
to that trendline rather than decrease as
50:19
it so often has done. So
50:21
what should the average church leader
50:25
and the average business leader, because we have a lot
50:27
of business leaders listening to... Yeah. So
50:30
what questions should they be asking? What should they
50:32
be paying attention to? What should
50:34
they be doing right now? Well,
50:37
the first thing goes for either if you're in an enterprise
50:39
context or you're in a small business or
50:41
if you're a church leader or in a
50:44
charity is do you even know what AI
50:46
is being used in your organization? If you
50:48
can't answer that question confidently,
50:51
then that's the first place to start. And that might
50:53
mean, yes, you need to do a survey and find
50:55
out how many people have signed up to a chat
50:57
TBT Plus with their private credit card and they're charging
51:00
it back. Or how many
51:02
of you have turned on an AI
51:04
module inside of your conferencing software without
51:06
looking at how that data is being
51:08
collected or stored? Do you
51:10
know how the tools that you're using
51:13
are using AI and what models are they using? Because
51:16
we've seen, as we've said, AI has shown up
51:18
now not just in a, hey, I bought an
51:20
AI product. It's showing up in all of the
51:22
technology we use from the mailbox provider that you
51:24
choose to things like Canva and Notion and Asana
51:27
and all of the things that we use to
51:29
keep our day-to-day projects going
51:31
in the digital ecosystem and many of those being
51:33
used, obviously, in the church as well. These
51:36
all are beginning to have AI tools added to
51:38
them. It may not be the headline. It may
51:40
be just a part of the stack. And
51:42
so the first thing to do is pay attention to that.
51:45
What's going on? Do you know what's being
51:47
used? The second is actually to use some of it and
51:49
crucially to pay for some of it, which I know
51:51
might seem counterintuitive. But what I mean
51:53
by that is use some of these tools to get
51:55
familiar with them. As I say, I don't think open
51:57
AI, what they're trying to do is evil. endorse
52:00
you, go play around with complexity and chat GPT,
52:02
but do it in a responsible way where you're
52:04
keeping things private that should stay private. Don't put
52:06
anything on the internet that you don't want on
52:08
the internet. It applies to AI tools as
52:10
well. And comply with all of
52:12
the usual things like GDPR and copper
52:15
compliance if you're in the US or whatever else
52:18
regulations apply at the time of listening
52:20
to this. Take
52:22
those regular precautions, but go and play around. And why
52:24
I say pay for it is because when you begin
52:26
to pay for these tools, they
52:28
are not using your time and attention as
52:31
the currency. They're using your money as
52:33
the currency in the interaction
52:35
with these models. And that means that more
52:38
often than not, you're not having any of
52:40
your data go to train future versions of
52:42
the model. You've got the
52:44
ability usually to delete what you've put in
52:46
and that you can remove it if you want to.
52:49
And that you will be getting enterprise level
52:51
safety and security at both ends when
52:53
you're sending data up and receiving it.
52:56
I did not know that. When you pay
52:58
for these things, you get those benefits. So
53:00
when you pay, you're under a stricter
53:03
level of privacy, security, etc. than when you're using
53:05
the free stuff. Did not know that. Just think
53:07
the difference between YouTube and YouTube Premium. What do
53:09
you get when you pay for YouTube Premium? You
53:12
don't get any ads. Because you're
53:14
no longer the product. I
53:16
happily pay for YouTube Premium in the past few years.
53:18
I do. That's all my nine I pay every month.
53:21
12.99 a month. Sign me up. But
53:24
it goes the same way. So when you pay
53:26
for Perplexity Plus or Pro, whatever they call it,
53:28
they're all called Pro Plus Max or Ultra, I
53:31
think is usually the way. But
53:33
whatever one you're paying for, the likelihood is
53:35
that the terms of service have changed and
53:37
you're no longer paying to train the model.
53:40
Your data is yours. So
53:42
again, you might not want to pay money to these people and
53:44
that's absolutely fine. But if you're going to use them, you're actually
53:46
better off paying for it most of the time. So
53:49
I would say first thing is know what's
53:51
going on. Second is try. Test it out.
53:54
And then the third thing is be transparent. Which
53:56
is if you're going to use these tools, whether
53:58
it's in the creation of... ministry resources, whether
54:00
it's in the administration of your business,
54:03
whether it's managing the data of your
54:05
people in whatever context, just
54:07
be upfront and transparent with the people that are
54:09
going to be affected by that. The best thing
54:11
you can do is publish ethics policy or a
54:13
transparency policy around AI, put it up on your
54:16
website in the same way you would do with
54:18
your privacy policy or other things, and
54:20
just state how you're using tools, what you're
54:22
using them for, crucially what you're not using
54:24
them for, and be upfront with
54:26
your congregation or your business about it because the
54:28
last thing that anyone wants is for there to be
54:31
some kind of sneaking suspicion that, hey, I think
54:33
the vicar or the pastor or the minister is
54:35
writing all of his statements with chat GPT and
54:38
it turns out to be true. You know,
54:40
at least if he's upfront about it then maybe
54:42
that's okay, but you don't
54:44
want to be found out and that's what
54:46
we want to make sure that people are
54:49
taken seriously. I think there's a lot of
54:51
fear around transparency
54:54
and I wonder if it's tied. I mean
54:56
you're working with the Church of England on
54:58
ethics on this. I just
55:01
read a case to, it might have been in the New
55:03
York Times, they have
55:05
an ethicist column and this pastor
55:07
served for 50 years, retired,
55:10
and someone in the church
55:12
just happenstance discovered another
55:14
sermon that sounded an awful lot like one
55:16
of his and then went back through his
55:18
work and discovered that they
55:21
were all ripped off, right down to the
55:23
stories that never happened to him. Like
55:25
told someone else's story as though he
55:27
had had that vacation or been to
55:29
that coffee shop, which to
55:32
me is unconscionable. The
55:34
ethicist was actually quite gentle saying, oh, let
55:37
this guy go into retirement. I'm like, no,
55:39
he lied. And yet
55:41
we've all used Google search for years,
55:43
right? I mean we use Google and
55:45
I mean... Anthosauruses
55:48
and concordances and assistant guides and
55:50
all these other things as well.
55:52
Biblical commentaries, it's all digitized now.
55:55
I still use chat DPT
55:57
for first drafts research. I've never found... a
55:59
good sermon to come out of AI yet.
56:02
That's just me. But sometimes if
56:04
I need something summarized or I want
56:06
ideas rebuilt or tested, I will
56:09
use it. So when it comes to
56:11
disclosure, like it's one thing to get up
56:13
there and say every Sunday, I used Google,
56:15
I read five commentaries, I went back to
56:17
the interlinear Greek, nobody does that. But I
56:19
have preached messages where I said, hey, a
56:21
lot of these ideas in the message, and
56:23
I'll give credit where they're due, came from
56:25
a sermon I heard from Tim Keller, or
56:28
Andy Stanley preached the series. I'm reteaching it right
56:30
now and it's got my own spin on it,
56:32
but you need to know these ideas came first
56:35
from Andy. I think that's a great example.
56:37
I think that's a great example. What
56:39
level of transparency would you give if
56:42
you weren't on the internet? That
56:45
should be the same level of transparency you
56:47
give with the internet, whether it's AI or
56:49
just Google. That's fine. That's what we're asking
56:52
for because it's
56:54
a good principle in general. I think it's a
56:56
Christian principle in general to be transparent about
56:59
these things, not necessarily to the
57:01
nth degree. Like you say, here's
57:03
my entire web search history that
57:05
went into this. That's not necessary. But
57:07
at least in the spirit of it is saying, hey,
57:09
I used... Or maybe it's once
57:12
a year or is part of your statement that's
57:14
just up on the website somewhere or on social
57:16
media that says, just so you know, these are
57:18
the tools that I regularly use to help with
57:20
my research and planning of this thing. It
57:23
hasn't happened to me yet where anything has
57:25
come out of Clodr, chat GPT or other
57:27
models where I'm like, oh, this
57:29
is like final draft 101. It's like, okay,
57:31
I have a few ideas. Let's get moving.
57:35
But I suppose the day is coming where
57:37
chat GPT will create a better sermon than
57:39
I could write or Clodr would.
57:42
I wonder... You're a good preacher. There's
57:44
a lot of like sermons that chat
57:46
GPT right now that's better than a
57:48
lot of sermons being written by humans.
57:50
Let's be honest. Perhaps.
57:52
And I guess you're at the point where
57:54
if I borrowed a John Orpburg sermon, I'm
57:56
just going to be like, hey, this started
57:59
with John Orpburg. or Berg, or,
58:02
and, you know, I put my spin on it,
58:04
or maybe you stand up one day and go,
58:06
you know what, chat GPT spit out this amazing
58:08
treatise on grace. And today
58:10
I'm going to share and a lot of
58:12
it was done via chat GPT. If it's
58:15
like 90% that and 10% you, I'm trying
58:17
to like, for example, AI helped write me
58:19
these questions for you. And I almost always
58:21
for the last year and a half, run
58:25
questions and guests through chat
58:27
GPT or Claude. And
58:29
then I sit down and I go, Oh, I never would
58:31
have thought to ask that. So it's
58:34
always a mix. It's somewhere between, sometimes
58:36
half. And sometimes it's like,
58:39
I kept one question or whatever.
58:42
So it's just the idea of transparency. And I
58:44
think you're right. Having a policy is really good.
58:46
Do you have a downloadable one? I know Kenny
58:49
Jang and church.tech and people like that are working
58:51
on that right now. I have not replicated one
58:53
because I would endorse the one that Kenny has.
58:55
So yeah, definitely go to church.tech and download his.
58:58
I was with him last week because we
59:00
were discussing exactly this and it's this is
59:02
the thing that I would recommend. So definitely
59:04
going if you don't follow Kenny, his AI
59:07
for church leaders, Facebook group is amazing. Church.tech
59:09
is great. So definitely go and
59:11
use that in the podcast. I recommend all his stuff.
59:14
But I would say that, you know, that's where we
59:17
do need to kind of, you know, we're getting more
59:19
of these practical guidelines, right? We're getting practical guidelines that
59:21
is beginning to emerge. I think the
59:23
theology of this stuff is still something we're going
59:25
to wrestle with for some time because,
59:28
you know, as these models emerge, what
59:30
happens when a great Christian theology based,
59:33
you know, model arrives that can write
59:35
sermons that was based on hopefully people
59:37
that opted into giving all of their
59:41
sermon content over. It
59:43
can create stuff. And then the question
59:45
is, well, where is, what is that? Is
59:50
that spoken into by God? Can
59:53
he use that stuff? And my answer
59:55
would be yes. I think God can use all things. That's
59:58
what we believe. But
1:00:01
particularly when it comes to teaching and
1:00:03
preaching, we have this very
1:00:06
specific call of ministry
1:00:08
onto that thing. And
1:00:10
we lump that in with the job of a church
1:00:12
leader. And I think
1:00:14
that's where we probably have gone a little bit
1:00:17
wrong along the way as we say the
1:00:19
church... For most church leaders, listening to this, if
1:00:22
that's your job, you probably are a staff member of
1:00:24
one. We often
1:00:26
think of the mega churches and the big places and that's
1:00:28
what a lot of the technology that's out there is catered
1:00:30
to. But for most people, certainly
1:00:32
here in the UK, the 17,000 churches across the Church
1:00:34
of England that I get to work with from time
1:00:37
to time, they're being led by
1:00:39
one person who doesn't have time to do this
1:00:41
stuff. And probably is trying to do all
1:00:43
aspects of leading a church and Kerry, I know you know
1:00:45
this firsthand, and not all of them are great
1:00:47
preachers. In fact, some of
1:00:49
them are really not great preachers at all. And
1:00:51
they would be the first to admit it because
1:00:53
they didn't get into leading ministry to be a
1:00:56
great preacher. They got into it, to
1:00:58
teach the poor or to be evangelists or to be
1:01:00
missioned. There's lots of different reasons. And
1:01:02
I think when
1:01:04
we think about these tools, there maybe is a
1:01:06
possibility that they could come along and really help
1:01:09
many people encounter the gospel in a way that they
1:01:11
weren't able to before through the voice of the person
1:01:14
in the local church who was speaking to them. But
1:01:16
we need to work through some of these things in terms
1:01:19
of like, what is our theology of that? And will we
1:01:21
accept that that can be part of the future of what
1:01:23
ministry looks like without necessarily going to
1:01:25
the nth degree of saying, we'll just stick chat
1:01:27
GPT with a voice assistant at the front of
1:01:29
church every Sunday and leave them to it. Well,
1:01:31
you bring your heart to it. You know, there's
1:01:33
a Canadian doctrine, I was just doing a quick
1:01:36
legal search from back in law school. It's a,
1:01:38
I don't think it's common law. So it could
1:01:40
be in England too. It's called passing off. And
1:01:42
it doesn't mean this, but in my head, I
1:01:44
always take that term and
1:01:46
apply it to church world. If I take,
1:01:49
you know, someone else's sermon or someone else's
1:01:51
book, like I've got William Uries here on
1:01:53
my desk, and I'm like reading
1:01:55
you chapter one and I pretend it's my
1:01:57
work. I'm passing it off as
1:02:00
though I did all the work, I did
1:02:02
all this stuff. And I think that's out
1:02:04
of bounds. I don't care whether it's analog
1:02:06
or digital. If you're pretending to do something
1:02:09
that you didn't do, on the other hand,
1:02:12
great books have great footnotes. And Tim
1:02:14
Keller joked that's been told on this
1:02:16
podcast once or twice, he was famous
1:02:18
for saying, well, first of all, almost
1:02:21
any random sermon you listen to by Tim Keller
1:02:23
probably has 25 to 50 references in
1:02:26
it. As C.S. Lewis says, as
1:02:28
Miroslav Volf says, as this person says, as
1:02:30
this person says, and you
1:02:32
don't think anything of it, you just think, wow,
1:02:34
that guy reads widely, which he does. And he
1:02:36
says, when I have time to write an original
1:02:38
sermon, that's how it goes. When I'm out of
1:02:40
time, I just quote C.S. Lewis. But he always
1:02:42
quoted C.S. Lewis. He never passed off. He would
1:02:45
say, as C.S. Lewis said, as C.S. Lewis said,
1:02:47
it was sort of a joke. But
1:02:49
one of the most profound thinkers of
1:02:51
this generation, Tim Keller, and he was
1:02:54
quoting people left, right, and center. So
1:02:56
if you're using chat GPT, throw your
1:02:58
heart into it, but just say,
1:03:00
hey, I had some help from this, or I had some
1:03:02
help here. But I think as a
1:03:04
pastor, you have to, and as a leader, I was
1:03:06
reading an article a friend sent me from
1:03:09
the Wall Street Journal, and it was
1:03:11
saying, what's happening to tech, Axios is
1:03:13
a media organization. And they're saying we're
1:03:16
humanizing the news at a very deep
1:03:18
level. They were founded in 2017
1:03:21
from people, I think, who started Politico, and
1:03:23
they're going to humanizing this at a very
1:03:25
deep level. I think when you bring your
1:03:27
humanity to the task, when I bring my
1:03:29
humanity to the show, when it's actually me
1:03:31
asking you the questions, and as you note,
1:03:33
like most guests do, how many
1:03:36
actual questions I sent you, did I actually use?
1:03:38
Very few. Because we're
1:03:40
having a great conversation.
1:03:42
But any agree, disagree? I'd
1:03:44
love your thoughts on that. Well,
1:03:47
no, so I agree. I think that
1:03:49
we all have to weigh
1:03:51
ourselves as to how much of my own
1:03:53
humanity do I want to bring to any
1:03:55
of these things that we do with AI,
1:03:57
whether that's preaching a sermon, whether that's writing
1:03:59
the annual. your address, whether it's writing a
1:04:02
birthday poem for a friend, whatever it
1:04:04
might be, or writing your... hopefully
1:04:06
you're not writing your wife's Christmas card,
1:04:08
kind of like a missive using AI.
1:04:11
But you might. You can
1:04:13
do all these things. We have to bring our humanity to and weigh
1:04:15
each of these things up individually. I mean, Kerry,
1:04:17
I'll turn it on you. How would you feel about
1:04:19
the idea of a Kerry bot in
1:04:21
the future? Long after you've kind of gone
1:04:23
to be with the Lord and
1:04:25
all of your sermon content falls over
1:04:27
into the public domain. Do you want
1:04:30
people interacting with Kerry 3.0 in the
1:04:32
future and downloading
1:04:35
future versions of sermons that you never wrote?
1:04:37
Write me a sermon on 1 Corinthians 13
1:04:39
with the theology of
1:04:42
C.S. Lewis and the style of Kerry Newhoff.
1:04:44
What would that look like? Yeah, I find
1:04:47
that very problematic. First of all, we were
1:04:49
talking about this as a team very recently.
1:04:51
I think most of the work I've done
1:04:53
in this life is going to disappear in a
1:04:55
generation. There's
1:04:57
very few people who get read beyond
1:05:00
their demise. But I think it...
1:05:02
Honestly, it's a great question, JP. Unless
1:05:06
things get upended very, very quickly,
1:05:08
I think the one thing
1:05:10
that has the potential to outlive you is a
1:05:12
book you write. Podcasts
1:05:15
come and go, blog posts come and
1:05:17
go, social media posts vaporize
1:05:20
with the dawn. I
1:05:22
think it's a book. And
1:05:25
what I like about a book when they're well
1:05:27
done is a book's taken you
1:05:30
anywhere from a lifetime to
1:05:32
multiple years to produce.
1:05:34
It's your best thinking on an issue. It's 30,
1:05:36
40, 50,
1:05:39
70,000 words on a dedicated subject.
1:05:42
And it makes a long-form argument that
1:05:45
you just can't make on
1:05:47
digital means. You can't do
1:05:49
it. And so, alternatively, if
1:05:53
you have a lecture series, you're a professor, it probably
1:05:56
has that level of thought to it in
1:05:58
lecture notes. But often, lecture in juror when
1:06:00
they get published. So I think if
1:06:02
you have a tome of work that
1:06:05
might be, by the time I'm gone, hopefully
1:06:07
I've written eight or nine books and that's
1:06:09
plenty for a lifetime. Maybe one of them
1:06:11
has a chance to be helpful beyond my
1:06:14
death, but the sermon chatbot
1:06:16
for Charles Spurgeon, I'm sure
1:06:19
is coming if it hasn't been invented yet.
1:06:21
But I think if you have legacy thinkers
1:06:23
or people who really helped shape the thought
1:06:26
of a generation that perhaps
1:06:28
a chatbot would be helpful. I
1:06:30
think for most of us, myself
1:06:32
included, it's like you made a contribution
1:06:34
in your lifetime, you did a good job and
1:06:37
most people tend, Gordon McDonald told me
1:06:39
years ago and he's written some amazing
1:06:41
books. I was asking him advice as
1:06:43
I stepped back from the teaching team
1:06:45
and like retired, retired at Connexus and
1:06:48
just became a congregant. He
1:06:50
said, Carrie, they forget you quickly. And
1:06:52
he was so accurate. And
1:06:54
that's humbling, but you just have to realize your
1:06:56
place in the culture. And so
1:06:59
yeah, I wouldn't have any doubt. I think
1:07:01
while I'm alive, we're working on a Carrie
1:07:03
bot so that people can ask
1:07:05
me questions and spit out, I have
1:07:07
thousands of articles and ideas
1:07:10
and published sources out there. We'll feed all
1:07:12
that into the GPT and
1:07:15
it'll pump out or LLM, it'll pump
1:07:17
out hopefully accurate answers. We haven't been
1:07:19
able to get it accurate enough yet,
1:07:21
but when we do get it,
1:07:23
I don't want to be responsible for saying things I
1:07:25
didn't mean and never intended to say. I
1:07:28
look forward to playing around with it. I might even write my
1:07:30
own seminar too using. That'd be
1:07:32
great. That'd be great. So I'm going to ask
1:07:34
you one or two more questions and then open
1:07:36
the floor to you. What do
1:07:39
you wish churches were doing with AI that they're not
1:07:41
doing right now? Wow.
1:07:43
I mean, there is a lot
1:07:45
just to kind of get off the starting blocks with,
1:07:47
which is again, we have
1:07:50
this wonderful opportunity, don't we, like in
1:07:52
church ministry to bring people
1:07:55
into relationship with Jesus and have them
1:07:57
grow and be nurtured and be discipled.
1:08:00
But the thing that often gets in the way
1:08:02
of that is just the day-to-day drudgery of managing
1:08:04
our money,
1:08:06
managing our teams, putting people on rotors,
1:08:08
all of this stuff. So one of
1:08:10
the big things I just wish churches
1:08:12
would be doing is seeing this massive
1:08:14
opportunity that's in front of them to
1:08:16
pick up a bunch of things that
1:08:18
might just make that task a little
1:08:21
bit easier this week, this month. Think
1:08:23
about, particularly in the post-digital age and
1:08:25
post-COVID, we're all wrestling with having to
1:08:27
basically compete for attention on social media,
1:08:29
to create content, to bring people through the
1:08:31
doors. It's the primary place
1:08:33
that people are spending time and money in terms
1:08:35
of evangelism and outreach. But most
1:08:38
churches don't have a church staff. As I
1:08:40
said before, most churches are one person with
1:08:43
an army of well-meaning volunteers. And
1:08:45
so they don't have teams that can help do
1:08:47
the social media posting or manage a schedule or
1:08:49
put together a rotor. And if
1:08:51
they are doing it, they're often having to do it in
1:08:54
their spare time. So I
1:08:56
think one of the big things is
1:08:58
that churches would adopt some of these
1:09:00
things just to get the basics done
1:09:02
and more of their well-earned time and
1:09:04
their congregation's resources on that, rather
1:09:06
than having to relearn how
1:09:08
to do another three social media posts in
1:09:10
Canva next Sunday. Robert Leonard So
1:09:12
you know what's fascinating? I've got a personal story on
1:09:14
that. We had some family over a couple of weeks
1:09:16
ago, and I have a niece who started her own
1:09:19
business. And she wasn't at the table,
1:09:21
but her mom was. And she was just saying
1:09:23
how overwhelming it was for my niece to
1:09:26
try to come up with social media. But she
1:09:28
wants to get the word out. And I'm like,
1:09:30
well, have you used AI? So we literally just
1:09:32
brought my laptop to the kitchen counter. And I
1:09:35
used a couple of AIs and just entered
1:09:37
a prompt that said, here's the name of
1:09:39
the salon. Here's the name of the person,
1:09:42
this location, this is specialization. Create
1:09:44
a social media plan for
1:09:46
the next 30 days on her Instagram
1:09:48
account. Literally, strategy,
1:09:51
exact text, everything,
1:09:53
image suggestions in
1:09:55
about 25 seconds. And my
1:09:57
sister-in-law started to cry. She's
1:10:00
like, this exists? I'm like,
1:10:02
absolutely. And that was the, well, I
1:10:04
think I was paid. I've done the same thing
1:10:06
with a teacher that I know, a friend of
1:10:08
mine, who's a teacher. She teaches fourth grade here
1:10:11
in the UK. And I showed
1:10:13
her how to transcribe a YouTube video, turn
1:10:15
it into a quick, short
1:10:17
story analogy for her class, and
1:10:19
then write a bunch of follow-up questions. And
1:10:23
that same reaction, just people's eyes light
1:10:25
up and like, this stuff is actually here. The
1:10:28
thing that I really want people to do is not just use it, but
1:10:30
know how to use it really well. Like, that's
1:10:32
where the real game changer is. So, yeah, I
1:10:34
don't know if you did this with your example,
1:10:37
with your friend, but, you know, telling it to
1:10:39
not just write social media policy, but tell it,
1:10:41
hey, you're going to behave like an absolute social
1:10:43
media expert. You know everything that you need to
1:10:45
know about how to post on this platform. You've
1:10:47
learned from all of the big influences. And
1:10:50
now this is all the information I want
1:10:52
you to go and produce for me. Those
1:10:54
little tweaks in the prompt do amazing things.
1:10:56
They still work. Okay, that's good to know.
1:10:58
They still work there. And these
1:11:00
models, they produce some fascinating results when you give
1:11:03
them a bit of personality to respond in before
1:11:05
you get started. So I think that's the other
1:11:07
thing I want people to do is use them
1:11:09
well. Well, there's
1:11:11
a lot of stuff we could get to, but we've
1:11:14
been an hour on this already. Anything
1:11:17
we haven't touched on that you'd like to touch
1:11:19
on? I think the main
1:11:21
thing I would say is that I
1:11:24
think that those are speaking to anyone that's
1:11:26
kind of working in technology or working in
1:11:28
leadership right now has a
1:11:30
responsibility when it comes to AI that they may not
1:11:32
think about, which is that
1:11:35
we are the last generation,
1:11:37
maybe the most anxious, we're
1:11:39
the last generation of
1:11:42
people, of leaders that will
1:11:44
remember what life was like before and
1:11:48
after this technology has
1:11:50
existed. Like my kids,
1:11:52
I've got at the time recording a
1:11:54
nine-year-old and a six-year-old, they
1:11:56
will not know what life is like to go
1:11:58
through school, to go into university. to
1:12:00
work in a workplace, to talk
1:12:03
to a social media thing and
1:12:06
not question whether or not is this a real person
1:12:08
or is this an AI? We're
1:12:11
the last generation to remember what that's like. And so
1:12:13
I think as leaders,
1:12:16
as technologists, as particularly
1:12:18
people working in ministry, we're
1:12:20
the last people to remember what that truly
1:12:22
disconnected experience was like. And we want to
1:12:25
try and look for what was the gold
1:12:27
in that that we want to bring into
1:12:29
the next generation and hold on to. And
1:12:32
what are the things that we want our AIs,
1:12:34
we want our technology to look like in the
1:12:36
future so that we end
1:12:38
up in that more positive view of
1:12:40
where that future curve might go rather
1:12:42
than the more negative. Because I think
1:12:45
we only get one shot at it.
1:12:48
We've seen so many times before that when this
1:12:50
technology or any technology gets out of the box,
1:12:52
there is no real putting it back in there. And
1:12:55
we're seeing obviously efforts at the moment to do that with
1:12:57
social media. We don't want to have to
1:12:59
go through the same thing with AI. And
1:13:01
I think that's our responsibility. It's all of
1:13:03
our responsibility. It comes back to the point
1:13:05
I made at the start, which is you
1:13:07
can wait for the government to regulate a
1:13:09
model or you can hope
1:13:12
that a benevolent AI council arises
1:13:14
out of the big seven companies
1:13:16
running the show or
1:13:19
you and your neighbors and your
1:13:21
children and the people that you work
1:13:23
with take collective responsibility to say, this
1:13:25
is what we want this to look
1:13:27
like in the future and
1:13:29
behave accordingly. And not just
1:13:31
give in to the easy thing,
1:13:34
not to give into the lazy thing
1:13:37
and not just give up more
1:13:39
utility for our privacy and end
1:13:41
up with unintended consequences. That's what
1:13:43
I really want people to try
1:13:45
and do. I think that is
1:13:47
a fantastic call to action for
1:13:49
older listeners. So I got
1:13:52
the internet when I was 31. I
1:13:54
got my first smartphone when I was 40 and
1:13:57
it was a blackberry. And I was so excited that I
1:13:59
could get emails on. It was primitive,
1:14:01
Philistine. I miss my black Yeah,
1:14:05
I know. Again, you're too young. But you
1:14:07
know what is interesting? And I've
1:14:09
often thought about pre-digital memory, but I was really
1:14:11
thinking about it. You know, I'm not young as
1:14:14
a leader anymore. And it's like, but
1:14:16
I do know what life was like before
1:14:19
we all got the internet, the prototype
1:14:21
of the internet, the dial up internet.
1:14:24
And I think, you
1:14:26
know, if you think about what it means to
1:14:28
be human, I have an interest in philosophy. It's
1:14:30
people like Voltaire. It's people like Sautre. It's people
1:14:33
like Kierkegaard and Nietzsche to some
1:14:35
extent, although he's pretty dystopian, who
1:14:40
really help us think... The most positive outlook.
1:14:42
Not the most positive outlook. And theologians who make
1:14:44
us think about what life is really about,
1:14:46
it would be wonderful in
1:14:49
these next few decades while those of us
1:14:51
who remember that are still alive, because we
1:14:53
won't be in 30, 40 years,
1:14:56
to know what it means to be
1:14:58
human and to really think about that.
1:15:01
Because I think you're right. Like civilization is about
1:15:03
to change fundamentally and
1:15:06
inalienably forever. And
1:15:09
maybe one of our contributions, because a lot of
1:15:11
older leaders feel useless. They feel like, oh, my
1:15:13
time is gone. I'm not a young leader anymore.
1:15:15
I think to be able to reflect
1:15:18
and prayerfully think through the theological, philosophical
1:15:20
dimensions of what was it like to be
1:15:23
the last generation to grow up with
1:15:25
a pre-digital memory and to interact
1:15:27
as human beings and
1:15:30
to have meaning in your
1:15:32
life. What does that look like? And
1:15:34
then the next generation will separate the wheat from the
1:15:36
chaff and figure out what it keeps and what
1:15:38
it discards. But that
1:15:40
is a tremendous help, I think, to people who
1:15:42
are older. And don't think you have to write
1:15:44
a book. Just get together with
1:15:47
some 25-year-olds and have conversations. That's
1:15:49
a beautiful call to action. I
1:15:52
think the thing that I hold onto is that
1:15:55
God could have chose to create anything.
1:16:00
have chosen to create us
1:16:02
as silicon-based intelligences, but he
1:16:04
chose to build us as carbon ones. And
1:16:07
I often have this phrase that I say a lot
1:16:09
which is that matter matters. We
1:16:11
are physical 3D living our
1:16:14
life out in the dimension of time for
1:16:16
a reason, because that's the way that God
1:16:18
intended it. And we don't
1:16:20
want to lose sight of that, that we get
1:16:22
a lot of these kind of digital fantasies that
1:16:24
live inside of the Silicon Valley bubble, thinking
1:16:26
that the future is us all being downloaded
1:16:29
onto a chip and having endless
1:16:31
life in that way. And that's just not
1:16:33
the model that we've been given. And I think we want
1:16:35
to hang on to that as best
1:16:37
as we can and be the story that
1:16:40
speaks into society that matter matters and that
1:16:42
we are built to be with one another
1:16:45
in an embodied way because that's... He
1:16:48
came to earth in bodily form for
1:16:50
a reason. He
1:16:53
didn't speak to us through a chatbot. Mm-hmm.
1:16:56
Fascinating. JP, people are going to want to
1:16:58
track with you. Where will they find you
1:17:00
online these days and any resources you want
1:17:02
to direct people to in particular? Yeah. Well,
1:17:05
I'm sure you'll find the links to the show notes, but
1:17:07
you'll find me at James Poulter at pretty much every social
1:17:09
media platform you need to. I write
1:17:12
mostly over on LinkedIn, if that's your place or if
1:17:14
you like to hang out. And
1:17:16
if you are interested, I
1:17:18
suppose, in getting more information
1:17:20
about how to follow along
1:17:22
with what we're doing at
1:17:24
Ecclesii, go to ecclesii.org. That's
1:17:26
A-E-C-L-E-S-I. That's the
1:17:29
Ecclesii.A-I-Bit.org. ecclesii.org.
1:17:31
And yeah, you'll get the resources, follow
1:17:33
the research, and engage in the conversation.
1:17:36
Fantastic. Thank you so much. Thank
1:17:38
you, Kerry. Man, I'm so glad
1:17:40
that we got to have this conversation.
1:17:42
Spent some time with JP recently in
1:17:45
London as well. He's a great leader
1:17:47
and I hope that was both practical
1:17:49
and kind of metaphysical for you.
1:17:51
I think that's what AI is
1:17:53
bringing to the table. Hey, next episode,
1:17:55
we've got Ken Blanchard and Randy Connelly.
1:17:57
Very excited to talk to the legend.
1:18:00
of the one-minute manager. We'll get the
1:18:02
backstory on that. Seagull management, the power
1:18:04
of simplicity and brevity, and a whole
1:18:06
lot more. And make sure you
1:18:09
check out today's partners. Did
1:18:11
you know I'm doing my first live
1:18:13
event? And if you haven't registered yet,
1:18:15
time is creeping up and
1:18:17
the spots are selling out fast.
1:18:19
So you can secure your spot
1:18:21
by going to the artofleadershiplive.com. That's
1:18:23
the artofleadershiplive.com. Join me for three
1:18:25
days in Dallas. I'm so excited
1:18:27
to do this. It's an intimate
1:18:29
event. I would love to have
1:18:32
you there. A long-time listener. I'm
1:18:34
talking to you. Go to theartofleadershiplive.com.
1:18:36
And our friends at 10 by
1:18:38
10 want to help you
1:18:40
figure out the impact that your
1:18:42
ministry is making with the next
1:18:44
generation. If you're a youth pastor,
1:18:47
fill out their free relational discipleship
1:18:49
inventory that will measure your youth
1:18:51
ministry's efforts. Go to 10 by
1:18:54
10.org/RDI today for your
1:18:57
free assessment. That's t-e-n-x-1-0.org/RDI
1:18:59
to learn
1:19:02
more. As I said, Ken Blanchard's
1:19:05
coming up. We've also got Priscilla
1:19:07
Schreyer, Rich Vilodis, Steve Cuss, Nikki
1:19:09
Gumbel, Max Locato is
1:19:11
coming back, Andy Stanley, Ed Stetser, and a
1:19:13
whole lot more. And T Wright, did I
1:19:16
mention him as well? And then I've
1:19:18
got something for you because you listened to the end. First
1:19:21
of all, if you enjoyed this, please leave a
1:19:23
rating and review. And secondly, make sure you check
1:19:25
out the Preaching Cheat Sheet. You know, over 10,000
1:19:27
leaders use it pretty much
1:19:29
every week to help them determine ahead of
1:19:31
time whether their message is going to land
1:19:33
or whether it needs a little more work.
1:19:35
So you can go to preachingcheatsheet.com, download your
1:19:37
free copy of it. The link is also
1:19:39
available in the show notes. And thank you
1:19:41
so much for listening, everybody. I appreciate you
1:19:43
so much. I have spent a lot of
1:19:45
time on the road and I got to tell you, I
1:19:48
have loved meeting so many of you. A lot
1:19:50
of you go back to episode one, which is
1:19:52
incredible because we've been at this almost a decade.
1:19:54
And man, you're giving me like real time feedback
1:19:57
on all the things that made a difference in
1:19:59
your life. and I appreciate you and I
1:20:01
appreciate that so much. I can't believe we get
1:20:03
to do this. All right. Well, I
1:20:05
hope today's episode helped you identify and break
1:20:08
a growth barrier you're facing. We'll catch you
1:20:10
next time on the podcast.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More