November 4, 2024 • 31 mins
Artificial intelligence is revolutionizing crisis communications, but what happens when powerful tools fall into the wrong hands? Jonathan Bernstein, chairman and founder of Bernstein Crisis Communications, joins host Peter Woolfolk to explore this complex landscape. You’ll hear about AI's ability to transform anyone into a seemingly credible source, raising pressing concerns about authenticity and the spread of misinformation. From empowering small PR firms with efficient press releases to posing ethical dilemmas in educational settings, AI's dual-edged nature is laid bare.
As AI tools like ChatGPT become ubiquitous, their applications in both professional arenas and activist circles are scrutinized. The rapid compilation of data can arm activists with information, true or false, against corporations. This episode tackles the challenges of discerning the truth in this AI-driven age, stressing the significance of mastering AI research skills. We also address the looming threat of AI-generated visuals and audio, which could easily mislead audiences if not carefully monitored.
The podcast doesn't just highlight the dangers but also emphasizes solutions. We discuss the urgent need for PR professionals to either attain AI expertise or collaborate with those who do. Jonathan Bernstein offers a candid warning: real-world experience is vital, and credentials should never be taken at face value. As the legal and ethical implications of AI-generated content continue to evolve, staying informed and vigilant is indispensable for anyone in the communication field. Join us as we unearth the potential and pitfalls of AI in public relations, and thank Jonathan Bernstein for sharing his invaluable insights.
Information on NEW podcast website.
Real Talk About MarketingAn Acxiom podcast where we discuss marketing made better, bringing you real...
Listen on: Apple Podcasts Spotify
Newsletter link:
https://www.publicrelationsreviewpodcast.com
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Announcer (00:03):
Welcome to the Public Relations Review Podcast, a
program to discuss the manyfacets of public relations with
seasoned professionals,educators, authors and others.
Now here is your host, peterWoolfolk.
Peter Woolfolk (00:23):
Welcome to the Public Relations Review Podcast
and to our listeners all acrossAmerica and around the world.
Now, this podcast has beenranked by Apple as among one of
the top 1% of podcasts worldwide, so I'd like to thank all of
our guests and listeners forbeing the basis for making this
happen, and we certainly wouldappreciate getting a review from
(00:45):
you at the end of the podcastas well.
Now, today, the topic for todayis the episode is crisis
communications, particularly asit applies to artificial
intelligence.
First, I'd like to read severalaccolades that my guest has
received for his work.
The first one says being agreat crisis manager is like
(01:05):
learning a new language orrunning a marathon.
We'd all love the accolades,but the amount of work it takes
to get there is overwhelming.
Even worse, the only way to getthe mastery is through an
excruciatingly painful process.
The second one says if you oryour organization do not wish to
suffer from the either deservedor undeserved criticism from
(01:27):
your actions or inactions, it'sincumbent upon you to call upon
Bernstein Crisis Management now.
But while they've reliablydemonstrated time after time
that can help clients minimizedamage after things have gone
bad, they really earn their paywhen they're given the
opportunity to show clients howto take proactive steps to avoid
(01:48):
problems before they occur.
Having worked in the crisismanagement field for most of my
professional life, I canhonestly say that Jonathan is
one of the best practitioners onthe planet.
So to seek his good counsel isan indication of your own
management powers, for it showsthat you understand that
protecting your organizationinvolves much more than hoping
(02:12):
nothing happens.
And this comes from a directorof public affairs at the
Import-Export Bank of the UnitedStates.
Would you be comfortablegetting crisis communication
from someone who's earned acertificate in crisis
communication by attending aseminar for several hours in a
hotel, or would you like to signup with a proven, battle-tested
(02:33):
expert, especially in the ageof artificial intelligence?
So my guest today is JonathanBernstein.
He is the chairman and founderof Bernstein Crisis
Communications.
He joins me today from PalmDesert, California.
So, Jonathan, welcome to thepodcast.
Thank you so much for having meLook.
As I mentioned, obviously,crisis communications is very
(02:57):
important to understandingproblems that can be caused by
ne'er-do-wells people who wantto cause some issues.
So what, your point of view,are some of the main things that
people should look out for whenit comes to artificial
intelligence as a damage that itcan possibly cause?
Jonathan Bernstein (03:15):
Well, peter, I can say definitively, after
some of my experiences of thelast year in particular, that
artificial intelligence is goingto test the human intelligence
of its users, and I've neverbeen so happy that I've been a
nerd for over 40 years and kindof grown up with the internet,
because it really helpsunderstand what systems like
(03:38):
ChatGPT are doing.
But you've already mentionedone of the big things that's
going to happen is that badactors will be empowered.
You've already mentioned one ofthe big things that's going to
happen is that bad actors willbe empowered For better or for
worse.
Ai turns incompetent writersinto moderately skilled writers
and competent speakers intomoderately skilled readers.
It can literally, visually orin writing, turn them into
(04:03):
something which presents assomething that they are not.
But it doesn't matter, becauseAI will change their image,
change their words, change theinflection speed of their voice
and any one of a number of otherfactors.
Normally you have to train ahuman being to do, and so it
(04:25):
ends up seemingly addingcredibility to communicators who
don't deserve it.
And I mean, on the plus sideit's you know, many ways.
For example, smaller PRoperations will benefit more, I
think, than larger ones,particularly those that have no
PR staff or agency orindividuals who have a complaint
because they can generate pressreleases.
(04:47):
I've had Chatsheet PD, forexample, draft a press release
for a mine disaster, simulate it.
It was a great press release.
I could improve it a little bit, but it was a great press
release.
And you know, unfortunatelycollege professors are seeing
young would-be professionals usethe same tools to generate
(05:10):
their homework and it's almostimpossible to detect other than
AI is sometimes better than theoriginal student and the
professor can tell that.
But I personally know twoprofessors who've gotten out of
the field because they're sodisgusted with what students can
do with AI.
So, and the thing that mostpeople, I think, forget about
(05:34):
with AI, but it's been true forthe lifetime of the modern
Internet, there was an oldacronym in the origin of the
Internet GIGO G-I-G-O.
Garbage in, garbage out.
Peter Woolfolk (05:45):
Oh yeah, I remember that one.
Jonathan Bernstein (05:47):
It's only, as you know, the use of AI is
only as good as the operator isskilled in making requests.
You know anybody can producethe basic results, but if you
really want good results, justlike with a good Google search,
you need to know how to do it.
And that's what's missing a lot.
I think staying on top of thelatest potential AI tools will
(06:12):
be challenging for anybody and,as with the early days of social
media, a lot of people won'tget it.
They just won't get it.
They will never click with that.
But you know, it still gives aversatility, tremendous
versatility across a wide rangeof media.
Graphics to illustrate news isvery easy to generate without
(06:34):
needing to buy stock images.
Unfortunately, or fortunately,depending on who you're with.
If you're a crisis firm whodeals with intellectual property
, well, there's going to be alot of allegations of
intellectual property theftcoming out of the way that AI
brings together information froma lot of different sources
without necessarilyacknowledging the source really
(07:08):
aware of.
Is that AI results?
If they're, for example, chat,gpt, are rife with inaccuracies
based on what the algorithmperceives to be the consensus
truth and that's a reallyimportant phrase.
I made that up.
I think it's still an importantphrase the consensus truth,
which isn't necessarily accurate.
So you have to fact check yourresults more than probably any
(07:28):
other search that's ever beendone, because it literally tries
to please you in some ways andmake up stuff that's totally
wrong, and I've caught it manytimes doing that.
That said, it's still fasterand more effective than a team
of young PR pros doing theresearch for you.
Peter Woolfolk (08:00):
Well, let me just answer that right quick,
because sometimes I certainly douse ChatGPT.
If I want to, let's say, maybedo a synopsis, for example, of
an opening letter, I'll let itwrite what it writes and then
I'll go through and then makethe changes that I feel are
necessary to make me feelcomfortable or put it in the
right way.
I'm just not, I just don'taccept what it presents to me.
It just presents, let's say,the skeleton, if you will, that
I'm looking for and I can go inand make the changes that I want
(08:21):
so that it fits my style andway that I communicate.
For all practical purposes itjust minimizes some of the time
I have to put into some issuesor writing.
Jonathan Bernstein (08:33):
Exactly.
It will also imitate styles.
So if you would like something,for example, to be written in
the style of William Shakespeare, it'll do that.
So if you want to sound more,you know, kamala Harris or
Donald Trump-like, you can dothat.
(08:53):
Clearly, donald Trump, stuffwasn't written by artificial
intelligence, because there'svery little intelligence.
But that's my opinion, butthat's my political bias, but
you know.
That said, the way that you canmake things fit as you say,
(09:19):
your style or anybody else'sstyle, is instantly I mean
instantly, literally in under 60seconds, usually with a chat
GPT is phenomenal, and I don'tthink people even realize the
versatility.
If you don't mind, let me tellyou some of the chat, gpt chats.
(09:40):
I've done research, I've donerecently.
Peter Woolfolk (09:43):
Okay.
Jonathan Bernstein (09:44):
Just to see the wide variety of things you
can do with this.
So I'm having a dispute with adentist right now.
So I've had a really goodresearch report prepared for me
on dental ethics and specific tothe state of California, and
gave me good information I cantake to my dentist and say here,
(10:06):
I don't think you're complyingwith this.
If I'm a big-time troublemakerfor a large corporation, I can
use the same tactics Look up allthe negative information you
can find on XYZ Corporation.
It'll find it all for me inseconds and then, whether it's
true or not, if I'm an activist,I might just publish the whole
(10:26):
thing just as it is, withouteven citing a source, since
nobody seems to care much aboutsources these days.
And this is the biggest risk itsounds credible.
Whether it is or not, it soundscredible and that's everything
on the internet is credible notbeing not actually being fact
checkable, but sounding likeyou're credible, and one of the
(10:49):
ways that activists used toundermine themselves in the past
was to rant a lot.
This allows them to get aroundthat problem and because ranting
undermines your credibilities.
But I've also done research onhow to to customize.
I wanted to make a customizedwrapping paper design.
It did that for me.
(11:09):
It looked up the origin of someimportant phrases, looked up
the importance of the pH balanceand cleaning carpets, gave me a
list of the top small businessinsurers so I could get out a
new liability policy.
You know, it gave me a lot ofelection night watch parties.
(11:31):
I wanted to know where therewere election night watch
parties, which resulted in myannouncing my own election night
watch party and literally youknow.
And where can I get what hotelsin Las Vegas feature adjoining
rooms?
I mean, you can ask anythingand you can ask for it in as
much detail as you want.
Announcer (11:51):
I want 10 examples.
Jonathan Bernstein (11:52):
I want 20 examples of hotel rooms.
I want 10 examples of what thiscompany has gotten in trouble
with, legally or not, and thegood thing is, if a client hires
me now, I can look for you know, I always look up data about a
prospective client immediatelyonline.
(12:12):
I'm always quick and good atGoogle searching, but now, if I
do the same search with ChatGPT,I get a lot more information a
lot more quickly, and solidinformation is critical to
crisis communications decisionmaking, and so this allows a lot
of information to be collated,sorted, organized in any way I
(12:34):
want, and I had a clientenvironmentally related clients
who was being accused ofgreenwashing, you know, claiming
to be green in their practicesbut actually not, and so I
wanted to understand all aboutgreenwashing practices that
they're being criticized for,and ChatGPT told me all about it
(12:56):
instantly.
So the uses are truly infinite,and I'm, you know, nerd though
I am, I'm still only barelytouching the surface of what can
be done done, but I haveexperimented both with images
(13:16):
and image related AI and withresearch related AI like chat
GPT.
I still haven't found anythingbetter than chat GPT.
Peter Woolfolk (13:20):
All the let's try the competition you know,
one of the things that I'mgetting out of this is that,
particularly when you say thatpeople need to read and check
what they're getting, fact-checkwhat they're getting, because
that is one of the problems thatcrisis communications is, or
part of crisis communications isthat these ChatGPT and several
(13:42):
of the others can writemisinformation and it's being
published and if you don't takethe time to read it and
double-check it, then you canperhaps have a problem,
particularly if you're trying toget some business or prevent
some business or use it for thebasis of taking some action, but
the information is incorrect,then you can create some
problems for yourself becauseyou didn't fact check what it is
(14:05):
that you got from thisparticular source.
Jonathan Bernstein (14:09):
Right, and hopefully good actors fact check
and bad actors don't, becausethen you can call them on it.
But, um, but you know, itreally depends on it's.
It's.
Peter Woolfolk (14:18):
It comes down to the skill of the operator, and
I I can't think of a moreimportant skill for a young vr
professional right now than tobecome a master of of ai
research well, other thing tooas we talked about the written
part, there's also some visualproblems that have come up, that
have been shown on the news,where real people appear to be
(14:41):
saying things that you wouldn'texpect them to say, so that
artificial intelligence canactually or some programs can
actually replicate voices ofreal people but put the
incorrect information into thosevoices, which really causes
some problems if people don'ttake the time to look into it.
Jonathan Bernstein (15:01):
Yes, and everybody knows the term
Photoshopped by now.
Well, think, photoshopped onsteroids.
That's AI.
Photoshopped is still a humanbeing making the changes, but AI
can make changes to images in away far superior to most human
(15:21):
beings and certainly quickerthan any human being, probably.
So, yes, I agree with you.
It's going to give theappearance of credibility or
truth to information that is notcredible or true, and most
people don't check, so badactors will get away with it,
(15:42):
and it'll be up to the peopleresponding to them to call out
the lies.
Peter Woolfolk (15:48):
Let me ask you then are there programs?
I mean, obviously there areprograms that can write whatever
you tell it to write inaccurateor whatever but are there
programs that can detectmisinformation?
Let's say that you seesomething and you're not quite
certain about it.
Is there a program that you canput this into and ask for it to
check for its accuracy?
For its accuracy?
(16:09):
Unfortunately no.
Jonathan Bernstein (16:16):
You've got.
When it comes to newsinformation, of course you know,
thank God there are now finallya lot of news organizations
doing fact-checking, but thatisn't all information.
So no, there isn't.
It's one of the things plaguingcollege professors.
It's very, you know, unlessthey want to personally pull out
data to try to find where thatbrilliant phrase came from over,
(16:45):
say, an entire thesis, and, forthat matter, even AI will give
you the sources.
It just may not be accurate.
A simple example when I askedfor information on hotels, it
gave me some links that are nolonger functional links.
I asked for URLs, but the linksaren't valid URLs anymore Now
(17:09):
if.
I was publishing that in atravel article I'd be giving
away information.
That's not that people wouldclick on it and go to a 404 page
error or, worse, to a bad site.
And yet that was produced by AI.
Peter Woolfolk (17:26):
It gets a little bit scary when you see that
these things are coming alongacross so many different
platforms.
You know, particularly onlineyou read almost every other day
that some governments or otherpeople are particularly now I
see that.
You know, particularly with thevoting coming up, that a lot of
(17:47):
misinformation is being passedalong and folks are not quite
sure what to do about it, how togo about checking it, maybe
other than calling up what mightbe the implied source about
that sort of thing.
Jonathan Bernstein (18:02):
Exactly, and AI is also assisting hackers
with human engineering you knowsocial engineering to get them
to appear to be someone theyweren't.
There was a famous case lastyear, maybe two years ago, that
the person responsible forcutting checks for a large
(18:25):
organization thought that theyhad received video calls and
audio calls from several membersof their executive team
authorizing them to redistributemoney to this place and that
place and this place.
It was all fake and theytransferred millions of dollars
(18:46):
outside the company and it wasall AI-generated fake.
Peter Woolfolk (18:59):
Well, you know, obviously this is a huge problem
and because it's so new, thereprobably are not ways that
people particularly if you'rejust an ordinary citizen going
through your day knowing what todo to double check it unless it
appears to come from somereliable source and you can get
in touch with that source tovalidate it.
But outside of that, there justdoes not seem to be a way that
people can double check it otherthan call some other people and
(19:23):
ask them for help of some kind.
Jonathan Bernstein (19:28):
No, unfortunately, peter, the AI
tools to generate newinformation are developing much
more quickly than any counteringforce.
It kind of again like happenedin the early days of viruses and
antivirus software.
Antivirus software used to havea really hard time keeping up
(19:49):
with all the viri that hackerswould invent.
It's gotten a lot better overthe years, but I think we're
kind of at the same stage withAI right now.
We're going to be behind theeight ball for a while on any
means of quickly fact-checkinginformation that AI produces,
because either that or you'rejust going to get people
(20:13):
considering everything to beeither treating everything
that's true or everything that'sfalse, and that won't help
anybody either.
Peter Woolfolk (20:21):
Well, I guess the other thing that has to
happen, or should happen, isthat those people who are being
tricked, if you will, or theorganizations that are being
misrepresented, once they findout, then they have to obviously
take some initiative, not onlywant to correct it, but also let
the public know that this ismisinformation that they should
disregard and come directly tothe company, rather than
(20:45):
depending upon some externalsource for certain information.
Jonathan Bernstein (20:50):
I agree, and also it's going to be
interesting this is justoccurring to me actually it's
going to be interesting what theimpact is of AI and the subject
of defamation.
If someone right now probably,and I'm not a lawyer, but in the
(21:17):
business, I'm in, I have toknow defamation law pretty well
to avoid getting accused ofdefamation.
Defamation law requires firstthat you knew the information
was false at the time youreleased it and that it was done
with malice.
And that's malice, is thehardest part.
Intent is hard to prove veryhard to prove.
But that said, if you generatesomething from AI and publish it
(21:41):
and you don't fact check it, doyou have a responsibility?
If it's inaccurate or false,does that constitute defamation?
It's inaccurate or false?
Does that constitute defamation?
I think defamation lawyers aregoing to have fun with that one
because it's going to happen.
(22:01):
Somebody is going to say youknow that, that you know
so-and-so person committed thiscrime or did this, or had a
relationship with small children, or whatever it is.
It's horrible that they saidthey did and and here's our
research from ChatGPT.
And they just cut and pasteChatGPT right into a press
(22:21):
release, right into a blog post,right into X or wherever, and
it's out there and not 100%retrievable at that point.
So what is the liability ofsomeone for not bothering to
fact-check something known to befactually inaccurate on a
(22:44):
regular basis?
Peter Woolfolk (22:49):
So that says to me that people need to be on
their toes, otherwise they'llhave some good liability
insurance to be on the safe sideand the insurers are going to
be pulling their hair out tryingto figure out how to cover that
.
Jonathan Bernstein (23:07):
So it's a science expanding so quickly
that its uses and misuses haveonly barely been touched.
Announcer (23:16):
That's what.
Peter Woolfolk (23:17):
I'm afraid of.
I mean, I see new things justlike.
Every week something new iscoming up and it's getting to
proportions that you know wehadn't even thought about or
might not have been capable ofthinking about.
That that can't be done.
So, jonathan, do you have anyother good words that you can
impart to us about this?
Jonathan Bernstein (23:42):
If you don't get on top of it.
The biggest thing is if you'rein the PR field.
Even if you're not.
If you're in the PR field, havesomeone available to you who
knows this stuff and stays ontop of it.
If you're not in the PR field,get someone who understands this
stuff and not just one part ofit, but who looks at AI across
(24:03):
the board.
You know my specialty is crisismanagement, so I look at it
from that perspective.
But it's going to be criticalfor traditional business
promotion too.
It's going to be critical fortraditional PR to build the
value of a brand.
One big area I think it's goingto have a huge risk in is
(24:24):
investor relations, because whatappears on the internet can
change, for a publicly heldcompany can really make a big
difference for better or forworse and depending on what it
is, and even if it's false, if acompany is accused of doing
something right away, it takes ahit.
So you know that's the downside.
(24:48):
The good side is, if you wantto create a broad cushion of
goodwill to get you through morecrisis, ai will let you
generate the copy at least thecontent much more quickly and
much more regularly.
So that you know everybody'salways struggled over, you know,
(25:09):
creating two, three, four blogposts a month, for example.
Not a problem anymore.
Creating two, three, four blogposts a month, for example.
Peter Woolfolk (25:18):
Not a problem anymore.
You know, one of the things Isaid, as we sort of begin to get
to the end of this, is that if,in fact, part of your job at
your company is crisiscommunications, that you need to
get with someone who has theexperience.
The idea of taking a class, youknow you're sitting in a
classroom or a hotel ballroom,there for a matter of hours and
(25:41):
having somebody explain to youthe different types of crisis
communication here's some of thesteps you can take to remediate
the problem I think is a very,very short-sighted idea, because
some things just you can'tlearn in the class.
You have to learn by experience.
You know what to do if there'sa fire in your building.
That's one thing.
(26:02):
But if someone has, you know,something comes out in terms of
that's created by artificialintelligence.
You don't know what to dobecause your course didn't cover
that.
Somebody needs to at leastbegin to identify somebody that
can help them, that has had someexperience with problems when
it comes to artificialintelligence.
(26:23):
Otherwise they're going to havea major problem on their hands.
Jonathan Bernstein (26:28):
Yeah, I agree, and it's going to come
down to some of us will be readyfor that, some of us are ready
for that and some will not be,but that will sort itself out
very quickly.
People will learn who can do?
This and who can't, just likewe did?
I mean, I'm sure you'veencountered this too.
(26:50):
So many people claim expertisein search engine optimization.
10% actually know what they'redoing.
It's got to be the same waywith AI.
Peter Woolfolk (27:01):
You know, I actually very, very briefly.
I did a show several monthsback about some, I think some
folks out of the college in Utahlooked at PR companies that
said that they were versed incertain topics I forget exactly
what it was right now and theyinteracted with a lot of them.
(27:22):
Some of them actually did havethe experience.
Others just used it as amarketing tool.
Yes, you know we're great atdoing A, b, c and D, but they
really never done it, and so youneed to be careful about things
like that too.
Ask for credentials.
If you're looking for somebodythat helps you at a certain
thing, ask them for somecredentials that you then can go
(27:43):
and check to find out ifthey're valid.
Jonathan Bernstein (27:46):
Exactly.
Yeah, I mean you wouldn't hirea brain surgeon without looking
up their credentials and recorda little bit.
Crisis management is kind ofthe brain surgery of PR.
Peter Woolfolk (27:56):
Good Well, jonathan, you've been a great
guest, as always.
I just wonder if there are anyclosing remarks, anything that
you think we've missed that weshould hear about.
Jonathan Bernstein (28:06):
No, Peter, other than you know who was the
famous football coach who saidyou know, now is today, or
something like that.
George Allen, I think, hadsomething like that to say way
back here Now is the time.
This is.
My message is now, you know,this is a good time.
It's not too late yet to learna whole lot more about AI and
(28:27):
how it'll work with PR ingeneral and if you're in my area
in crisis management, Know whatit can do, know what it can't
do, and admit it if it doesn'tmake any sense to you, and find
someone to whom it does makesense.
Peter Woolfolk (28:43):
Absolutely Well.
Let me say again thank you toJonathan Bernstein.
He's the chairman and founderof Bernstein Crisis
Communications.
We certainly hope that you'veenjoyed his presentation today
and, of course, if you do, we'dlike to get a review from you,
and we'd also like for you toshare this information with your
friends and colleagues, andalso for the next edition of the
(29:06):
Public Relations Review.
Join us, thank you.
Announcer (29:12):
This podcast is produced by Communication
Strategies, an award-winningpublic relations and public
affairs firm headquartered inNashville, Tennessee.
Thank you for joining us.
Welcome to the Public Relations Review Podcast, a
program to discuss the manyfacets of public relations with
seasoned professionals,educators, authors and others.
Now here is your host, peterWoolfolk.
Peter Woolfolk (00:23):
Welcome to the Public Relations Review Podcast
and to our listeners all acrossAmerica and around the world.
Now, this podcast has beenranked by Apple as among one of
the top 1% of podcasts worldwide, so I'd like to thank all of
our guests and listeners forbeing the basis for making this
happen, and we certainly wouldappreciate getting a review from
(00:45):
you at the end of the podcastas well.
Now, today, the topic for todayis the episode is crisis
communications, particularly asit applies to artificial
intelligence.
First, I'd like to read severalaccolades that my guest has
received for his work.
The first one says being agreat crisis manager is like
(01:05):
learning a new language orrunning a marathon.
We'd all love the accolades,but the amount of work it takes
to get there is overwhelming.
Even worse, the only way to getthe mastery is through an
excruciatingly painful process.
The second one says if you oryour organization do not wish to
suffer from the either deservedor undeserved criticism from
(01:27):
your actions or inactions, it'sincumbent upon you to call upon
Bernstein Crisis Management now.
But while they've reliablydemonstrated time after time
that can help clients minimizedamage after things have gone
bad, they really earn their paywhen they're given the
opportunity to show clients howto take proactive steps to avoid
(01:48):
problems before they occur.
Having worked in the crisismanagement field for most of my
professional life, I canhonestly say that Jonathan is
one of the best practitioners onthe planet.
So to seek his good counsel isan indication of your own
management powers, for it showsthat you understand that
protecting your organizationinvolves much more than hoping
(02:12):
nothing happens.
And this comes from a directorof public affairs at the
Import-Export Bank of the UnitedStates.
Would you be comfortablegetting crisis communication
from someone who's earned acertificate in crisis
communication by attending aseminar for several hours in a
hotel, or would you like to signup with a proven, battle-tested
(02:33):
expert, especially in the ageof artificial intelligence?
So my guest today is JonathanBernstein.
He is the chairman and founderof Bernstein Crisis
Communications.
He joins me today from PalmDesert, California.
So, Jonathan, welcome to thepodcast.
Thank you so much for having meLook.
As I mentioned, obviously,crisis communications is very
(02:57):
important to understandingproblems that can be caused by
ne'er-do-wells people who wantto cause some issues.
So what, your point of view,are some of the main things that
people should look out for whenit comes to artificial
intelligence as a damage that itcan possibly cause?
Jonathan Bernstein (03:15):
Well, peter, I can say definitively, after
some of my experiences of thelast year in particular, that
artificial intelligence is goingto test the human intelligence
of its users, and I've neverbeen so happy that I've been a
nerd for over 40 years and kindof grown up with the internet,
because it really helpsunderstand what systems like
(03:38):
ChatGPT are doing.
But you've already mentionedone of the big things that's
going to happen is that badactors will be empowered.
You've already mentioned one ofthe big things that's going to
happen is that bad actors willbe empowered For better or for
worse.
Ai turns incompetent writersinto moderately skilled writers
and competent speakers intomoderately skilled readers.
It can literally, visually orin writing, turn them into
(04:03):
something which presents assomething that they are not.
But it doesn't matter, becauseAI will change their image,
change their words, change theinflection speed of their voice
and any one of a number of otherfactors.
Normally you have to train ahuman being to do, and so it
(04:25):
ends up seemingly addingcredibility to communicators who
don't deserve it.
And I mean, on the plus sideit's you know, many ways.
For example, smaller PRoperations will benefit more, I
think, than larger ones,particularly those that have no
PR staff or agency orindividuals who have a complaint
because they can generate pressreleases.
(04:47):
I've had Chatsheet PD, forexample, draft a press release
for a mine disaster, simulate it.
It was a great press release.
I could improve it a little bit, but it was a great press
release.
And you know, unfortunatelycollege professors are seeing
young would-be professionals usethe same tools to generate
(05:10):
their homework and it's almostimpossible to detect other than
AI is sometimes better than theoriginal student and the
professor can tell that.
But I personally know twoprofessors who've gotten out of
the field because they're sodisgusted with what students can
do with AI.
So, and the thing that mostpeople, I think, forget about
(05:34):
with AI, but it's been true forthe lifetime of the modern
Internet, there was an oldacronym in the origin of the
Internet GIGO G-I-G-O.
Garbage in, garbage out.
Peter Woolfolk (05:45):
Oh yeah, I remember that one.
Jonathan Bernstein (05:47):
It's only, as you know, the use of AI is
only as good as the operator isskilled in making requests.
You know anybody can producethe basic results, but if you
really want good results, justlike with a good Google search,
you need to know how to do it.
And that's what's missing a lot.
I think staying on top of thelatest potential AI tools will
(06:12):
be challenging for anybody and,as with the early days of social
media, a lot of people won'tget it.
They just won't get it.
They will never click with that.
But you know, it still gives aversatility, tremendous
versatility across a wide rangeof media.
Graphics to illustrate news isvery easy to generate without
(06:34):
needing to buy stock images.
Unfortunately, or fortunately,depending on who you're with.
If you're a crisis firm whodeals with intellectual property
, well, there's going to be alot of allegations of
intellectual property theftcoming out of the way that AI
brings together information froma lot of different sources
without necessarilyacknowledging the source really
(07:08):
aware of.
Is that AI results?
If they're, for example, chat,gpt, are rife with inaccuracies
based on what the algorithmperceives to be the consensus
truth and that's a reallyimportant phrase.
I made that up.
I think it's still an importantphrase the consensus truth,
which isn't necessarily accurate.
So you have to fact check yourresults more than probably any
(07:28):
other search that's ever beendone, because it literally tries
to please you in some ways andmake up stuff that's totally
wrong, and I've caught it manytimes doing that.
That said, it's still fasterand more effective than a team
of young PR pros doing theresearch for you.
Peter Woolfolk (08:00):
Well, let me just answer that right quick,
because sometimes I certainly douse ChatGPT.
If I want to, let's say, maybedo a synopsis, for example, of
an opening letter, I'll let itwrite what it writes and then
I'll go through and then makethe changes that I feel are
necessary to make me feelcomfortable or put it in the
right way.
I'm just not, I just don'taccept what it presents to me.
It just presents, let's say,the skeleton, if you will, that
I'm looking for and I can go inand make the changes that I want
(08:21):
so that it fits my style andway that I communicate.
For all practical purposes itjust minimizes some of the time
I have to put into some issuesor writing.
Jonathan Bernstein (08:33):
Exactly.
It will also imitate styles.
So if you would like something,for example, to be written in
the style of William Shakespeare, it'll do that.
So if you want to sound more,you know, kamala Harris or
Donald Trump-like, you can dothat.
(08:53):
Clearly, donald Trump, stuffwasn't written by artificial
intelligence, because there'svery little intelligence.
But that's my opinion, butthat's my political bias, but
you know.
That said, the way that you canmake things fit as you say,
(09:19):
your style or anybody else'sstyle, is instantly I mean
instantly, literally in under 60seconds, usually with a chat
GPT is phenomenal, and I don'tthink people even realize the
versatility.
If you don't mind, let me tellyou some of the chat, gpt chats.
(09:40):
I've done research, I've donerecently.
Peter Woolfolk (09:43):
Okay.
Jonathan Bernstein (09:44):
Just to see the wide variety of things you
can do with this.
So I'm having a dispute with adentist right now.
So I've had a really goodresearch report prepared for me
on dental ethics and specific tothe state of California, and
gave me good information I cantake to my dentist and say here,
(10:06):
I don't think you're complyingwith this.
If I'm a big-time troublemakerfor a large corporation, I can
use the same tactics Look up allthe negative information you
can find on XYZ Corporation.
It'll find it all for me inseconds and then, whether it's
true or not, if I'm an activist,I might just publish the whole
(10:26):
thing just as it is, withouteven citing a source, since
nobody seems to care much aboutsources these days.
And this is the biggest risk itsounds credible.
Whether it is or not, it soundscredible and that's everything
on the internet is credible notbeing not actually being fact
checkable, but sounding likeyou're credible, and one of the
(10:49):
ways that activists used toundermine themselves in the past
was to rant a lot.
This allows them to get aroundthat problem and because ranting
undermines your credibilities.
But I've also done research onhow to to customize.
I wanted to make a customizedwrapping paper design.
It did that for me.
(11:09):
It looked up the origin of someimportant phrases, looked up
the importance of the pH balanceand cleaning carpets, gave me a
list of the top small businessinsurers so I could get out a
new liability policy.
You know, it gave me a lot ofelection night watch parties.
(11:31):
I wanted to know where therewere election night watch
parties, which resulted in myannouncing my own election night
watch party and literally youknow.
And where can I get what hotelsin Las Vegas feature adjoining
rooms?
I mean, you can ask anythingand you can ask for it in as
much detail as you want.
Announcer (11:51):
I want 10 examples.
Jonathan Bernstein (11:52):
I want 20 examples of hotel rooms.
I want 10 examples of what thiscompany has gotten in trouble
with, legally or not, and thegood thing is, if a client hires
me now, I can look for you know, I always look up data about a
prospective client immediatelyonline.
(12:12):
I'm always quick and good atGoogle searching, but now, if I
do the same search with ChatGPT,I get a lot more information a
lot more quickly, and solidinformation is critical to
crisis communications decisionmaking, and so this allows a lot
of information to be collated,sorted, organized in any way I
(12:34):
want, and I had a clientenvironmentally related clients
who was being accused ofgreenwashing, you know, claiming
to be green in their practicesbut actually not, and so I
wanted to understand all aboutgreenwashing practices that
they're being criticized for,and ChatGPT told me all about it
(12:56):
instantly.
So the uses are truly infinite,and I'm, you know, nerd though
I am, I'm still only barelytouching the surface of what can
be done done, but I haveexperimented both with images
(13:16):
and image related AI and withresearch related AI like chat
GPT.
I still haven't found anythingbetter than chat GPT.
Peter Woolfolk (13:20):
All the let's try the competition you know,
one of the things that I'mgetting out of this is that,
particularly when you say thatpeople need to read and check
what they're getting, fact-checkwhat they're getting, because
that is one of the problems thatcrisis communications is, or
part of crisis communications isthat these ChatGPT and several
(13:42):
of the others can writemisinformation and it's being
published and if you don't takethe time to read it and
double-check it, then you canperhaps have a problem,
particularly if you're trying toget some business or prevent
some business or use it for thebasis of taking some action, but
the information is incorrect,then you can create some
problems for yourself becauseyou didn't fact check what it is
(14:05):
that you got from thisparticular source.
Jonathan Bernstein (14:09):
Right, and hopefully good actors fact check
and bad actors don't, becausethen you can call them on it.
But, um, but you know, itreally depends on it's.
It's.
Peter Woolfolk (14:18):
It comes down to the skill of the operator, and
I I can't think of a moreimportant skill for a young vr
professional right now than tobecome a master of of ai
research well, other thing tooas we talked about the written
part, there's also some visualproblems that have come up, that
have been shown on the news,where real people appear to be
(14:41):
saying things that you wouldn'texpect them to say, so that
artificial intelligence canactually or some programs can
actually replicate voices ofreal people but put the
incorrect information into thosevoices, which really causes
some problems if people don'ttake the time to look into it.
Jonathan Bernstein (15:01):
Yes, and everybody knows the term
Photoshopped by now.
Well, think, photoshopped onsteroids.
That's AI.
Photoshopped is still a humanbeing making the changes, but AI
can make changes to images in away far superior to most human
(15:21):
beings and certainly quickerthan any human being, probably.
So, yes, I agree with you.
It's going to give theappearance of credibility or
truth to information that is notcredible or true, and most
people don't check, so badactors will get away with it,
(15:42):
and it'll be up to the peopleresponding to them to call out
the lies.
Peter Woolfolk (15:48):
Let me ask you then are there programs?
I mean, obviously there areprograms that can write whatever
you tell it to write inaccurateor whatever but are there
programs that can detectmisinformation?
Let's say that you seesomething and you're not quite
certain about it.
Is there a program that you canput this into and ask for it to
check for its accuracy?
For its accuracy?
(16:09):
Unfortunately no.
Jonathan Bernstein (16:16):
You've got.
When it comes to newsinformation, of course you know,
thank God there are now finallya lot of news organizations
doing fact-checking, but thatisn't all information.
So no, there isn't.
It's one of the things plaguingcollege professors.
It's very, you know, unlessthey want to personally pull out
data to try to find where thatbrilliant phrase came from over,
(16:45):
say, an entire thesis, and, forthat matter, even AI will give
you the sources.
It just may not be accurate.
A simple example when I askedfor information on hotels, it
gave me some links that are nolonger functional links.
I asked for URLs, but the linksaren't valid URLs anymore Now
(17:09):
if.
I was publishing that in atravel article I'd be giving
away information.
That's not that people wouldclick on it and go to a 404 page
error or, worse, to a bad site.
And yet that was produced by AI.
Peter Woolfolk (17:26):
It gets a little bit scary when you see that
these things are coming alongacross so many different
platforms.
You know, particularly onlineyou read almost every other day
that some governments or otherpeople are particularly now I
see that.
You know, particularly with thevoting coming up, that a lot of
(17:47):
misinformation is being passedalong and folks are not quite
sure what to do about it, how togo about checking it, maybe
other than calling up what mightbe the implied source about
that sort of thing.
Jonathan Bernstein (18:02):
Exactly, and AI is also assisting hackers
with human engineering you knowsocial engineering to get them
to appear to be someone theyweren't.
There was a famous case lastyear, maybe two years ago, that
the person responsible forcutting checks for a large
(18:25):
organization thought that theyhad received video calls and
audio calls from several membersof their executive team
authorizing them to redistributemoney to this place and that
place and this place.
It was all fake and theytransferred millions of dollars
(18:46):
outside the company and it wasall AI-generated fake.
Peter Woolfolk (18:59):
Well, you know, obviously this is a huge problem
and because it's so new, thereprobably are not ways that
people particularly if you'rejust an ordinary citizen going
through your day knowing what todo to double check it unless it
appears to come from somereliable source and you can get
in touch with that source tovalidate it.
But outside of that, there justdoes not seem to be a way that
people can double check it otherthan call some other people and
(19:23):
ask them for help of some kind.
Jonathan Bernstein (19:28):
No, unfortunately, peter, the AI
tools to generate newinformation are developing much
more quickly than any counteringforce.
It kind of again like happenedin the early days of viruses and
antivirus software.
Antivirus software used to havea really hard time keeping up
(19:49):
with all the viri that hackerswould invent.
It's gotten a lot better overthe years, but I think we're
kind of at the same stage withAI right now.
We're going to be behind theeight ball for a while on any
means of quickly fact-checkinginformation that AI produces,
because either that or you'rejust going to get people
(20:13):
considering everything to beeither treating everything
that's true or everything that'sfalse, and that won't help
anybody either.
Peter Woolfolk (20:21):
Well, I guess the other thing that has to
happen, or should happen, isthat those people who are being
tricked, if you will, or theorganizations that are being
misrepresented, once they findout, then they have to obviously
take some initiative, not onlywant to correct it, but also let
the public know that this ismisinformation that they should
disregard and come directly tothe company, rather than
(20:45):
depending upon some externalsource for certain information.
Jonathan Bernstein (20:50):
I agree, and also it's going to be
interesting this is justoccurring to me actually it's
going to be interesting what theimpact is of AI and the subject
of defamation.
If someone right now probably,and I'm not a lawyer, but in the
(21:17):
business, I'm in, I have toknow defamation law pretty well
to avoid getting accused ofdefamation.
Defamation law requires firstthat you knew the information
was false at the time youreleased it and that it was done
with malice.
And that's malice, is thehardest part.
Intent is hard to prove veryhard to prove.
But that said, if you generatesomething from AI and publish it
(21:41):
and you don't fact check it, doyou have a responsibility?
If it's inaccurate or false,does that constitute defamation?
It's inaccurate or false?
Does that constitute defamation?
I think defamation lawyers aregoing to have fun with that one
because it's going to happen.
(22:01):
Somebody is going to say youknow that, that you know
so-and-so person committed thiscrime or did this, or had a
relationship with small children, or whatever it is.
It's horrible that they saidthey did and and here's our
research from ChatGPT.
And they just cut and pasteChatGPT right into a press
(22:21):
release, right into a blog post,right into X or wherever, and
it's out there and not 100%retrievable at that point.
So what is the liability ofsomeone for not bothering to
fact-check something known to befactually inaccurate on a
(22:44):
regular basis?
Peter Woolfolk (22:49):
So that says to me that people need to be on
their toes, otherwise they'llhave some good liability
insurance to be on the safe sideand the insurers are going to
be pulling their hair out tryingto figure out how to cover that
.
Jonathan Bernstein (23:07):
So it's a science expanding so quickly
that its uses and misuses haveonly barely been touched.
Announcer (23:16):
That's what.
Peter Woolfolk (23:17):
I'm afraid of.
I mean, I see new things justlike.
Every week something new iscoming up and it's getting to
proportions that you know wehadn't even thought about or
might not have been capable ofthinking about.
That that can't be done.
So, jonathan, do you have anyother good words that you can
impart to us about this?
Jonathan Bernstein (23:42):
If you don't get on top of it.
The biggest thing is if you'rein the PR field.
Even if you're not.
If you're in the PR field, havesomeone available to you who
knows this stuff and stays ontop of it.
If you're not in the PR field,get someone who understands this
stuff and not just one part ofit, but who looks at AI across
(24:03):
the board.
You know my specialty is crisismanagement, so I look at it
from that perspective.
But it's going to be criticalfor traditional business
promotion too.
It's going to be critical fortraditional PR to build the
value of a brand.
One big area I think it's goingto have a huge risk in is
(24:24):
investor relations, because whatappears on the internet can
change, for a publicly heldcompany can really make a big
difference for better or forworse and depending on what it
is, and even if it's false, if acompany is accused of doing
something right away, it takes ahit.
So you know that's the downside.
(24:48):
The good side is, if you wantto create a broad cushion of
goodwill to get you through morecrisis, ai will let you
generate the copy at least thecontent much more quickly and
much more regularly.
So that you know everybody'salways struggled over, you know,
(25:09):
creating two, three, four blogposts a month, for example.
Not a problem anymore.
Creating two, three, four blogposts a month, for example.
Peter Woolfolk (25:18):
Not a problem anymore.
You know, one of the things Isaid, as we sort of begin to get
to the end of this, is that if,in fact, part of your job at
your company is crisiscommunications, that you need to
get with someone who has theexperience.
The idea of taking a class, youknow you're sitting in a
classroom or a hotel ballroom,there for a matter of hours and
(25:41):
having somebody explain to youthe different types of crisis
communication here's some of thesteps you can take to remediate
the problem I think is a very,very short-sighted idea, because
some things just you can'tlearn in the class.
You have to learn by experience.
You know what to do if there'sa fire in your building.
That's one thing.
(26:02):
But if someone has, you know,something comes out in terms of
that's created by artificialintelligence.
You don't know what to dobecause your course didn't cover
that.
Somebody needs to at leastbegin to identify somebody that
can help them, that has had someexperience with problems when
it comes to artificialintelligence.
(26:23):
Otherwise they're going to havea major problem on their hands.
Jonathan Bernstein (26:28):
Yeah, I agree, and it's going to come
down to some of us will be readyfor that, some of us are ready
for that and some will not be,but that will sort itself out
very quickly.
People will learn who can do?
This and who can't, just likewe did?
I mean, I'm sure you'veencountered this too.
(26:50):
So many people claim expertisein search engine optimization.
10% actually know what they'redoing.
It's got to be the same waywith AI.
Peter Woolfolk (27:01):
You know, I actually very, very briefly.
I did a show several monthsback about some, I think some
folks out of the college in Utahlooked at PR companies that
said that they were versed incertain topics I forget exactly
what it was right now and theyinteracted with a lot of them.
(27:22):
Some of them actually did havethe experience.
Others just used it as amarketing tool.
Yes, you know we're great atdoing A, b, c and D, but they
really never done it, and so youneed to be careful about things
like that too.
Ask for credentials.
If you're looking for somebodythat helps you at a certain
thing, ask them for somecredentials that you then can go
(27:43):
and check to find out ifthey're valid.
Jonathan Bernstein (27:46):
Exactly.
Yeah, I mean you wouldn't hirea brain surgeon without looking
up their credentials and recorda little bit.
Crisis management is kind ofthe brain surgery of PR.
Peter Woolfolk (27:56):
Good Well, jonathan, you've been a great
guest, as always.
I just wonder if there are anyclosing remarks, anything that
you think we've missed that weshould hear about.
Jonathan Bernstein (28:06):
No, Peter, other than you know who was the
famous football coach who saidyou know, now is today, or
something like that.
George Allen, I think, hadsomething like that to say way
back here Now is the time.
This is.
My message is now, you know,this is a good time.
It's not too late yet to learna whole lot more about AI and
(28:27):
how it'll work with PR ingeneral and if you're in my area
in crisis management, Know whatit can do, know what it can't
do, and admit it if it doesn'tmake any sense to you, and find
someone to whom it does makesense.
Peter Woolfolk (28:43):
Absolutely Well.
Let me say again thank you toJonathan Bernstein.
He's the chairman and founderof Bernstein Crisis
Communications.
We certainly hope that you'veenjoyed his presentation today
and, of course, if you do, we'dlike to get a review from you,
and we'd also like for you toshare this information with your
friends and colleagues, andalso for the next edition of the
(29:06):
Public Relations Review.
Join us, thank you.
Announcer (29:12):
This podcast is produced by Communication
Strategies, an award-winningpublic relations and public
affairs firm headquartered inNashville, Tennessee.
Thank you for joining us.
Popular Podcasts
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.