0:05

Grab a girl here. I'm in Jan Fogarty and

0:07

welcome to this bonus episode. There are words of

0:09

the year to talk about and I just could

0:12

not let 2023 go by

0:14

without talking about them. And

0:16

I'm here today with Jess Zafaris who joined us

0:18

recently to talk about her new book, Words from

0:20

Hell. And she is so kind to come back

0:22

and talk about words of the year with us.

0:25

Welcome Jess. Thank you

0:27

so much. You know, I love the show and I'm excited

0:29

to be back. Yeah.

0:32

And see, we are recording

0:34

this December 16th, but it won't come out

0:36

until I believe the 28th. So there could

0:38

be more words of the year that we

0:40

don't get to that come out after the

0:43

show, but at least you'll get the big

0:45

picture, some of them. So,

0:47

you know, I had been thinking about doing

0:50

this. And then when Jess, when you came

0:52

out with the Reagan words

0:54

of the year, I thought, I just can't let

0:56

the year go by without doing this. And you're

0:58

the perfect person to do it. So do

1:00

you want to start talking about Reagan's

1:03

word of the, Reagan communications word of

1:05

the year? Yeah, absolutely. So in addition

1:07

to writing about etymology for our listeners,

1:09

I'm also the director of content at

1:12

Reagan communications. So I made a list

1:14

of 10 words of

1:16

the year for Reagan, or

1:18

sorry, for Reagan's audiences. So

1:22

these words for were for

1:24

communications professionals and PR professionals

1:26

as well. And

1:28

as you well know, Mignon, because you lead

1:31

webinars for Reagan PR daily, communicators

1:33

are also writers and their relationship

1:35

with AI is learning

1:38

how to ethically and creatively incorporate AI

1:40

into their work streams to boost their

1:43

productivity and maintain

1:45

quality. So they they're

1:47

there, you know, that's been a key skill

1:49

set that they've been developing this year. And

1:51

a lot of the words of the year

1:53

we've seen have been focused on AI, as

1:56

I'm sure you've addressed. So the the

1:58

key area

2:00

for communicators is prompting

2:03

chatbots. So that was the number one

2:05

word of the year for Reagan and

2:07

PR Daily was prompt. And within that

2:09

umbrella, I would include skill sets, such as

2:12

training on AI or training and

2:14

AI on your style and voice,

2:16

building your own GPTs and other

2:19

adjacent skills. So

2:22

Reagan wasn't the only company to have prompt

2:25

this year either, were you? That's

2:27

right. We'll get into Oxford's word

2:29

of the year in a moment, which

2:31

is also AI adjacent, but one of

2:33

their runners up was prompt. And

2:37

although AI has factored into like PR

2:39

and comms in contexts like data management

2:41

and customer service chat, learning the art

2:44

of the prompt really became a requirement

2:46

for working in that field over the

2:48

past year. And obviously, I was not

2:50

as clever as I thought I was when

2:53

coming up with this term. But I

2:56

came up with it before I was

2:58

aware of Oxford's choice. And

3:00

part of the reason is it's a

3:02

little bit of a double entendre. Organizations

3:04

as a whole and communicators in particular

3:07

need to be prompt in upskilling in

3:10

this area to remain competitive. So

3:12

you weren't the only one who your

3:15

word was chosen by someone else.

3:17

This year has been, I mean,

3:19

as people will notice by the

3:21

end, almost every word this year

3:23

relates to AI in some way.

3:26

And AI was the actual word

3:28

chosen by at least two organizations.

3:30

Collins dictionary was the first one

3:32

to come out with AI, obviously

3:35

short for artificial intelligence. And

3:38

then the National Association

3:41

of National Advertisers chose

3:43

AI as the word of the year. And

3:46

they said it was the widest margin in

3:48

their 10 year history. For

3:50

that word, they vote. And so I think

3:54

the next runner up was something

3:57

like personal purpose, it was purpose.

3:59

And but it was just a distant

4:01

second to AI. And so,

4:03

you know, huge theme for the year.

4:05

And again, like multiple organizations

4:08

chose it. You know, it

4:10

doesn't surprise me that Purpose had such a

4:12

large gap. I feel like that was the

4:14

word of the year in the advertising space

4:16

like three years ago. Like brand Purpose was

4:18

the thing, the drum that they've been beating

4:20

for quite some time. And although AI isn't

4:23

new, it was absolutely the word you

4:25

got tired of hearing throughout this year.

4:28

Right. Yeah. I think you're going to

4:30

talk about hallucinate next. And I

4:33

had heard of people when I saw

4:35

it was announced, people said

4:37

they'd never heard it used in

4:39

relationship to AI. And I thought, I mean,

4:41

I couldn't go,

4:43

you know, two days without hearing it.

4:46

So we all live in different

4:48

environments. Absolutely. Yes, dictionary.com's

4:50

word of the year is AI themed,

4:53

but not one that you might expect.

4:55

They said that to determine their word

4:58

of the year, they gave themselves a

5:00

prompt using, there's that word again, prompt,

5:03

using lexicography and data science, choose

5:05

a single word that best represents

5:07

at this moment, AI's many profound

5:09

ramifications for the future of language

5:11

and life. And the result was

5:14

the word hallucinate, but a specific

5:16

definition of the word that

5:18

was just added to their dictionary this year. It's

5:21

specifically when an artificial intelligence produces

5:23

false information contrary to the intent

5:25

of the user and presents it

5:28

as if it is true and

5:30

factual, it's called hallucinating. And their

5:32

example sentence was, when

5:35

chat bot hallucinate, the result is

5:37

often not just inaccurate, but

5:39

completely fabricated. And I've

5:41

personally witnessed chat bot hallucinating in

5:43

my work. I asked chat

5:45

GPT, for instance, for the etymology of several

5:48

words that I knew to have common

5:50

myths associated with them in online

5:52

spaces. And sure enough, it

5:55

presented the myths as if they were

5:58

fact. So that's an example of hallucinating.

6:00

hallucination. And just just

6:02

to humor me a little bit, the origin

6:04

of hallucinate is one of my favorites. It's

6:07

from Latin and Greek

6:10

sources meaning to wander in the

6:12

mind and the

6:14

English word dates back to at least the 1630s

6:16

when it's recorded in a few texts with

6:19

the apparent meaning to like waver like

6:21

to hallucinate between truth and falsehood, which

6:23

I think is kind of a full

6:25

circle moment for today's definition that we

6:28

just that just earned word of the air. Definitely.

6:31

Oh, that's so cool. Yeah, I think

6:33

the first thing I ever wrote about

6:36

AI was about a hallucination about a

6:38

year ago, I had an experience similar

6:40

to yours. And I asked it to

6:42

recommend books that had librarians as a

6:45

protagonist. And it came up with,

6:47

you know, four or five that were great. And then

6:49

there was one that just did not exist. And

6:52

but it was by a real author. And,

6:54

you know, it might be something he would have written. But it just

6:56

wasn't. It

6:59

was not a real book. And, and so I wrote

7:01

up a blog post about

7:04

AI hallucinating and that. And

7:06

it's interesting too, because one of the, you know,

7:08

as I did in

7:10

the webinar, I did with with you

7:12

at Regan calm. So I

7:14

did the advanced AP style webinar. And because AI is

7:17

such a big thing this year, you had me include

7:19

a section on AI for AP, AP style. And one

7:21

of the things they emphasize is to, you know, I

7:23

think that's a really interesting thing. One of the things

7:26

they emphasize is to not

7:28

ascribe human characteristics to

7:30

AI like chat, GPT.

7:32

And I've always thought that hallucinate

7:35

sort of walks that

7:37

line because when you say that AI

7:39

has hallucinated in a way it feels

7:42

like you're ascribing a human characteristic

7:44

or condition at least to

7:47

that machine. And that's

7:49

the industry standard term. It's what

7:51

they use to talk about the

7:54

mistakes that the AI tools make.

7:56

It's a real industry term. Rather

7:59

than say. that it generates misinformation or

8:01

something along those lines, which is a little more

8:03

on the nose and a little less psychedelic,

8:06

shall we say. Right.

8:09

I love the graphic, but if

8:11

any of you have seen it

8:14

online, dictionary.com had a fabulous chaotic

8:16

graphic to go with, their word

8:18

hallucinate. Beautiful

8:21

piece of artwork, I agree. I'm interested

8:23

to see if next year's

8:25

words of beer will be AI

8:27

focused because like

8:29

next year, I think that the year will,

8:31

it will be a year of AI competitor

8:34

drama. Like there's

8:37

like ByteDance was using open AI

8:39

tech to build its own large

8:41

language model or LLM, which is like

8:44

the model that makes generative conversational AI

8:46

possible. Apparently

8:49

that's a pretty big faux pas and now they've

8:51

been banned. So that's been a thing. And then

8:54

there is Grock, which like

8:56

Bloomberg has an article that says it could

8:58

overtake chat GPT, but I doubt it. That's

9:01

Elon Musk's like subscriber

9:03

only AI chat bot that

9:06

was named after Heinlein's stranger

9:08

in a strange land and supposedly informed by

9:10

Hitchhiker's Guide to the Galaxy. But I, you

9:13

know, we'll see. It

9:15

also does not. It is like

9:18

a function of hallucination because it

9:20

professes it doesn't profess to present

9:22

information neutrally, but instead is designed

9:24

to answer with, quote unquote, a

9:27

bit of wit and a rebellious streak and

9:29

is useful to people of

9:31

all backgrounds and political views, which

9:33

just reads as like this is

9:35

probably going to be politically in a

9:37

certain direction in the context of Elon

9:39

Musk's diatribes. However,

9:43

you know, evidently someone put it to

9:45

the test and it ended up being

9:47

more like politically liberal than chat GPT

9:50

in many contexts. So

9:52

apparently they're working to try to make

9:54

it more conservative. So that'll be interesting.

9:57

It also it also it also played your

10:00

chat GPT which

10:02

is fascinating because this was apparently

10:04

experts predicted that happening wild

10:07

wild yeah well chat GPT itself

10:09

was one of the words of

10:11

the year so the Economist

10:13

and the Shanghai daily actually which

10:16

is I believe the only English

10:18

newspaper in Shanghai both chose chat

10:20

GPT which stands for chat generated

10:23

pre-trained transformer and yes I did

10:25

have to look that up and

10:27

you know

10:29

but I find myself wondering you know you were thinking next

10:31

year will I was wondering

10:34

if chat GPT will become like

10:36

Google if we'll use it as a verb

10:38

if the brand name will become so you

10:41

know genericized that we will say like

10:43

oh I searched chat GPT when we

10:46

were searching Grok or Claude or one

10:48

of the other ones I

10:50

kind of think not because I'm

10:52

not sure that chat GPT is

10:54

gonna win the

10:56

competitive race by that much

10:58

it's it's it's probably the best one out

11:00

there right now but I'm not sure if

11:03

it's gonna hang on to that crown it

11:05

was with

11:07

Wikipedia's most viewed page last

11:10

year to the chat GPT

11:12

page so you know

11:14

and it's the one that I think

11:16

of when I you know I started

11:18

as a side project I'm writing a

11:21

newsletter about AI called AI side quest

11:23

and you know it's just like little side project I'm

11:26

doing and every time I write about it I

11:28

should I refer to these tools as AI

11:31

should I refer to them as large language

11:33

models can I just call them chat you

11:35

know I know I can't call them all

11:37

chat GPT but I'm tempted to because that's

11:39

the name that everyone knows so you

11:41

know I feel tempted to use it in the

11:43

way you might say I googled something but

11:46

I don't and I resist that so I

11:49

think it'll be interesting to see what the next

11:51

year holds yeah you know I

11:53

think I think the way Google has

11:55

started incorporating AI into

11:57

its search

12:00

engine is almost more

12:03

natural in terms of actual fact

12:05

finding. And it's a little more

12:07

useful because it also includes the

12:10

sources from which it's drawing its

12:12

information. One thing that

12:14

I think could save ChatGPT if they

12:16

actually commit to it, if OpenAI decides

12:19

to commit to it, is that the

12:21

Data and Trust Alliance announced

12:23

data provenance standards intended to

12:25

provide clarity around the information

12:27

that forms the foundation of

12:30

AI generated works. So

12:32

basically food labels saying where

12:34

each piece of information came

12:36

from and what

12:38

recipe went into AI generated

12:40

text and images. So the hope

12:43

is that much like food safety

12:45

labeling that provides information about how

12:47

and where food is produced and

12:49

handled, the standards will provide users

12:52

with the ingredients and sources that

12:54

AI text or image generator is

12:56

used to develop their content, which

12:59

I like that a lot because it

13:02

lends authorship to the original

13:05

information and sourcing. Yeah, no,

13:07

that's great. A tool that I've

13:09

been using called Perplexity AI gives the

13:11

links to all the facts from where

13:13

it gets the information. And I

13:16

feel like I should caution people that even

13:18

when you see the sources there, you

13:21

need to check them because they

13:23

don't always support the fact that

13:25

the AI has given you. I

13:29

heard the same thing about Bing yesterday. So

13:31

it'll give you the sources, but those sources

13:33

might not support what it's saying it supports.

13:35

So if you're looking at facts, be

13:37

incredibly skeptical and click through on

13:39

those sources and check that they

13:42

really are or better

13:44

yet verify it yourself from some other

13:46

way. Don't trust the facts you get

13:48

from any of these large language models

13:51

slash chat GPT slash AI tools. I

13:54

would hope that people were doing that already with

13:56

sources like Wikipedia, but I'm pretty sure that

13:58

many people are not. Yeah,

14:00

and I think that one of the things

14:03

these tools like ChatGPT do is they give

14:05

you the answer in such a confident, clear

14:07

way that it's easy to believe

14:09

it in a way that is,

14:12

I think, more so than a lot of the

14:14

other ways that we get information online right now.

14:17

So don't be lured into

14:19

that feeling of, oh, this confident, friendly

14:21

friend of mine just told me a

14:24

fact that just it must be true.

14:26

It's so sure. Speaking

14:36

of friends, another word of

14:38

the year is authentic from

14:40

Merriam Webster. And

14:42

this comes from the Greek originally

14:45

that meant original or genuine. And

14:48

I think this was so interesting because

14:50

I remember authentic, you know,

14:52

Bob, wow, probably seven or eight years

14:55

ago, being a word that I was

14:57

sick to death of hearing. You

14:59

know, I think everyone was talking

15:01

about how millennials, you know, valued

15:03

authenticity and it was a big deal

15:05

years ago. And so at first I was surprised

15:08

when I saw that it was the

15:10

Merriam Webster word of the year. And then

15:12

I realized, again, it's in the context of AI. And

15:14

it has to do with determining whether a photo

15:17

or a piece of content

15:20

is authentic or fake. One

15:23

of their runners up was Deepfake,

15:25

actually. So, you know, we

15:27

had a lot of Deepfake images produced

15:29

by AI this year that people had

15:31

to deal with. One

15:33

cool thing that that gives me a little bit

15:36

of hope is some of the reactions that you

15:38

see, you know, as

15:40

quickly as AI is advancing. It's

15:43

also being used to combat some

15:45

of the things that were afraid of happening.

15:48

So, for example, in the

15:50

realm of Deepfakes, I believe it

15:52

was Intel developed a tool that

15:55

can that can determine whether

15:57

a video is a Deepfake.

16:00

fake by basically like

16:02

when you or I talk we can

16:04

only see each other's faces as they

16:06

are. But there are micro

16:08

changes in the colors of your skin because

16:10

there is blood going to different parts of

16:12

your face. So their

16:15

tool is able to determine whether it's whether a

16:17

video is a deep fake or not because a

16:19

deep fake will not have those

16:21

micro changes in color from blood

16:23

flow. Wild. I

16:26

have not heard that. I do. I

16:28

don't know the answer to this either,

16:30

but I wonder like to what degree

16:33

it's able like if it if

16:35

its accuracy is impacted by people

16:38

of different skin tones. Yeah,

16:41

I was just going you had reminded

16:43

me that I feel like I should also warn people

16:46

against the text

16:48

based AI identification

16:50

tools. So especially in a lot

16:53

of schools, they have

16:55

new tools that are claiming that they can

16:57

identify AI written essays,

16:59

for example, and they are

17:02

wildly inaccurate and they're

17:04

they're flagging things that aren't written by

17:06

AI quite often and

17:08

they are particularly susceptible to

17:10

misflagging essays written by people

17:13

for whom English isn't their

17:15

first language. So and

17:17

just something about the way the language is

17:19

a little bit different and these tools

17:21

are saying, oh, that was written by AI

17:23

when it wasn't and it's causing students all

17:25

sorts of problems and stress and so

17:28

be very, very cautious about using those

17:31

tools that claim they can identify AI

17:33

written text because they're not very good. Most of them

17:35

right now today, you know, maybe a year from now they

17:37

will be. But if you're listening to this

17:39

today, they're not the

17:42

world of complications that arises from

17:44

this technology, right? Yeah, so

17:46

many complications. And I think

17:48

that we had one

17:50

word of the year to wrap up

17:53

one word that wasn't AI focused that

17:56

you were going to talk about. That's

17:58

right. Oxford University Press selected one. Rizz

18:00

as its word of the air and

18:02

it's designed as style, charm or attractiveness,

18:04

the ability to attract a romantic or

18:07

sexual partner. And this actually does have

18:09

an AI tie in, I'll get to

18:11

that in a second. It

18:14

does. But it can,

18:16

so it can be a noun, I have

18:18

Rizz or a verb, like

18:20

you can Rizz someone up or attract them.

18:24

It's attributed in most

18:26

contexts to YouTuber Kai Sinat,

18:28

I hope I pronounced his

18:30

name right. He introduced

18:32

it in a video in 2021 and

18:35

a lot of people say it's short

18:37

for charisma, which makes sense. It

18:39

sort of like works in the same way that the

18:41

word like fridge comes out of refrigerator. But

18:44

Kai says that it's not, it's more

18:46

of a more of an inventive word.

18:49

So he also used phrases like W

18:52

Rizz or L Rizz basically winning

18:54

or losing at having Rizz. I

18:57

sound so old saying. But

19:00

Kai said he's a guy though. Right?

19:04

It's sort of like it reminds me a little bit about

19:06

how like jazz has been used in the past.

19:08

Like it's been, it has been several parts

19:10

of speech in the same way. But

19:12

Kai says he doesn't use it anymore because

19:15

TikTok butchered the word. It's

19:17

really common over there. You'll probably run into it

19:19

if you go on TikTok. It's

19:23

also been boosted by celebrities like Tom

19:25

Holland, who said he didn't have much

19:27

Rizz and had to play the long

19:29

game to date Zendaya. And

19:32

then it's also been used by public

19:34

figures like Scaramucci used it

19:37

to say that Rhonda Santis has

19:39

no Rizz, which I would generally

19:41

agree with. But I also

19:43

like listening to Rich Gen X public figures use

19:45

the phrase is to use another word from

19:47

my Zoomer friends, a little cringe. Definitely

19:50

cringe. So the AI tie

19:53

in that I was reading about just

19:55

yesterday, though, this isn't like specifically associated

19:57

with its word of the

19:59

year, but there is an AI tie

20:01

in. AI-powered dating assistant called Riz, which

20:03

suggests that, it suggests

20:06

conversation ideas for dating app users.

20:08

So it can come up with

20:10

your opening line on Tinder or

20:12

Bumble or whatever, and then you can

20:14

ask it what you should say next in response

20:16

to your potential partner's message. Oh

20:19

my gosh, so we are gonna have AI

20:21

chatbots talking to each other

20:23

in the chat on dating

20:26

apps. So if a person on

20:28

each side of a conversation has this, the

20:31

bots talking to each other. Right? That

20:35

reminds me of a story, I may not get the

20:37

details on this exactly right, so I'll try to be

20:39

kind of vague, but I remember a

20:41

story from several years ago

20:43

before chat GPT was a thing,

20:45

before any of this happened, when

20:48

they were experimenting with artificial

20:50

intelligence and machine learning, and they had

20:52

two bots

20:55

speaking to each other, just conversing.

20:57

And over time, they started

21:00

developing their own abbreviated language,

21:02

like they started abbreviating words

21:04

so much, and then making

21:06

their own dialect. It was

21:08

fascinating. Oh, I'm gonna

21:10

look that up and find out more about it. That

21:12

is fascinating. That's a good one. If I can find

21:14

it, I'll put a link to it in the show

21:16

notes, because that would be really cool to read about.

21:19

I'm sure people love that. It was a little creepy

21:21

too, but it was cool. Yeah. Awesome.

21:25

Well, thank you so much Jess

21:27

for helping us round up the words of

21:29

2023. Well,

21:32

to wrap up, thank you so much

21:34

for being here. I'm gonna talk about

21:36

our books. I have the grammar daily.

21:38

I'll put a link to my AI

21:40

newsletter in the show notes and to

21:42

my course with Regging Communications that Jess

21:44

helped put together. And Jess,

21:47

why don't you, thank you so much first, and

21:49

why don't you tell people about your book and

21:51

where they can find you? Thank

21:54

you so much. I just released

21:56

the book Words From Hell. It's

21:58

an etymology dictionary focused on nodding.

22:00

nefarious and secretly salacious words

22:02

among many others. You

22:06

can buy it anywhere books are sold and

22:08

then you can also find me on my

22:10

website at useless etymology and on tiktok at

22:12

jessafares. Thanks so much for having me on

22:14

the show. This has been so much fun. So

22:17

have a great new

22:19

year everyone. See you soon. Bye! That

22:26

was fun. I am loving your book, by the way. It is

22:28

so much fun. Thank you. I

22:30

loved yours too. Thank you. You

22:32

can clip that in if you want. Like an online

22:34

testimony. I

22:36

think it... Oh, I didn't stop recording. I thought I got...