00:00:00: They need a research and development budget to explore how AI can help their company, their industry.

00:00:08: If they don't have a budget in the team around that I think it's going to fall behind.

00:00:11: That is number one.

00:00:13: Impulses & Perspectives The DIKT Expert Talk hosted by Nikolaj Baer

00:00:21: Welcome another episode of impulses&perspective The DIkT Expert talk.

00:00:28: Today we are discussing a topic that is moving from science fiction into business reality faster than many leaders expected.

00:00:38: Deepfake's AI generated deception, digital trust and the future of leadership!

00:00:45: And my special guest today is Bryce Austin Cyber Security Advisor, former chief information security officer and co-author of our new English language book The Age Of Fakes.

00:00:59: How AI abuse fake news in deep fakes threaten business and society?

00:01:05: Bryce great to have you here!

00:01:08: Thank You very much Nikolai.

00:01:12: When did you personally realize that I generated deception was no longer just a theoretical problem, but the real problem for companies and organizations?

00:01:24: To be honest with you is when i saw how much better The fishing emails were getting that we're coming in both to me And my clients.

00:01:42: Significant holes in it.

00:01:44: Well, that doesn't happen anymore.

00:01:46: In fact I'll go up on stage now in front of an audience and take that email from twenty eighteen And put it into chat GPT and say make this sound like a professor From the University Of London and not quite so urgent and It is imperfect within ten seconds.

00:02:02: well i'm Not The only one doing That Is An Example.

00:02:05: There Are Cyber Criminals Around The Globe Doing That Every Single Day?

00:02:12: our employees to be on the lookout for, to know when it's a malicious email as opposed to legitimate one.

00:02:19: It doesn't work anymore —it is turning the world in its

00:02:21: head.".

00:02:23: Was there specific moment or technological development that made you think this was going change?

00:02:31: fundamentally how organizations have to operate?

00:02:35: That's good question!

00:02:38: I will say of AI.

00:02:42: to be able to manipulate the sort of information that both goes into and comes out in an organization is very strong.

00:02:51: And there are two concerns with them, if they're organizations not utilizing AI for their own benefit.

00:03:01: as example I used hire a developer.

00:03:04: he would write code help with vulnerability scans.

00:03:07: my company does enhance penetration tests.

00:03:10: Um, I know what i want out of a computer and I Know enough about them to know.

00:03:15: What is it isn't reasonable?

00:03:16: To ask but i'm not a coder.

00:03:17: i'm Not never been interested in that.

00:03:20: well i don't need to be anymore.

00:03:22: i can have A couple hour conversation with an ai engine And get what i need out Of it not have to hire developer any more.

00:03:30: Well there are companies That if they Don't understand where AI Can Have a competitive advantage They're going to miss out.

00:03:38: So I think that's a huge problem for companies, it kind of have their head in the sand.

00:03:43: And there are people using this against them not just from a cyber security angle because that is my specialty but to try and put out potentially negative press about other companies.

00:03:56: The fact AI can try to detect those things on social media Can be looking at what your company right now.

00:04:04: It's powerful tool.

00:04:06: So if companies aren't investigating AI right now, I think in five years they may very well be irrelevant.

00:04:13: Definitely definitely so.

00:04:15: probably you have the same observation when i do a keynote on these topics.

00:04:21: A lot of people Think still deepfakes are probably something like entertainment or an internet culture.

00:04:30: Why do you think business leaders underestimate that topic?

00:04:33: Still

00:04:36: changes hard.

00:04:37: you know, Kodak invented digital photography.

00:04:42: They did it so that we didn't have to wait for a satellite taking pictures around the globe To send of the film back down to earth and have it develop.

00:04:50: they wanted in real time And kodak came up with that technology But their executives got there bonuses based on how many developing chemicals?

00:04:59: That they sold.

00:05:00: and So the very technology that they invented put traditional photography essentially out of business, well they ignored it.

00:05:10: I think a lot of executives aren't ready for this level of change and its difficult to realize that what got us where we are isn't going get us the way need be.

00:05:20: so thats fundamental shift in not just how use technology but do work.

00:05:26: That's a big deal.

00:05:27: I don't know that we've seen it shift this fundamental since the internet kind of came to be useful.

00:05:33: maybe what, twenty five years ago around the year two thousand?

00:05:37: Yeah and i bought my first plane ticket on the Internet.

00:05:39: I think in maybe ninety four or ninety-five um... And I remember doing banking On The Internet at ninety seven ish when I moved to San Diego Um..and It was A Fundamental Shift.

00:05:49: You Know it Was Last Time you Were Inside Of Bank Unless You Needed Cash.

00:05:52: We Just Don't Do That Anymore.

00:05:54: AI is coming and it's going to change things quickly.

00:05:58: And the companies that aren't willing, we have to rethink how the world will work five years from now are gonna be left behind.

00:06:07: So what worries you more today?

00:06:09: The technology itself or this speed at which society adopts it.

00:06:16: From my standpoint in the cyber security area It worries me about the technology because its powerful.

00:06:24: I there is a project out called mythos right now that has trying to look for vulnerabilities in areas of code.

00:06:32: That have been around for years or even decades and it's doing very good job on.

00:06:36: i think the difficulty with AI is not perfect.

00:06:41: makes mistakes will get things wrong.

00:06:44: as cyber criminal attacker you don't care If it is only right in ninety percent of the time and ten percent at a time, it royally messes up whatever you're doing.

00:06:54: Well just chalk that to experience then move on another victim.

00:06:57: well as a defender we need be all the time an AI not write.

00:07:03: I just asked it last week.

00:07:06: I was watching a science fiction show with my family, and we wanted to know about the speed of light.

00:07:11: And how long?

00:07:11: It would take to get messages too and from Mars because in this show is hypothetical there.

00:07:17: so i ask chad gpt I knew The distance in the speed-of-light From the earth To the sun Was eight or nine minutes.

00:07:23: well How far as it's mars.

00:07:25: So i asked it and it came back three to twenty two Minutes and i'm like that's not possible.

00:07:30: i said mars does Not come inside the orbit Of the Earth.

00:07:33: And then it came back, oh I'm so sorry.

00:07:35: Yeah...I sent you the speed of light between Earth and Mars.

00:07:38: that's not what you asked.

00:07:39: You ask for sun to mars.

00:07:41: Like how did ya miss that?

00:07:43: That is pretty fundamental.

00:07:45: Well as a defender..that was horrific.

00:07:48: if i thought there were some terrible attack going on thats never happened.

00:07:52: Thats alot waste resources.

00:07:54: Or if they are obvious signs an attack AI just missed because its having bad day.

00:07:58: I can't tolerate that As an attacker, oops screwed that one up move on to another victim.

00:08:05: It worries me.

00:08:06: for the next three or five years Nikolai we are going to have a deluge of AI enhanced cyber attacks and AIs being used to defend it is but defending as different problem than attacking And Ai's know where near good at.

00:08:22: so thats my biggest fear.

00:08:24: For maybe even several year I think our cybersecurity problems are going to get worse, not better because of AI.

00:08:32: Absolutely agree and it's so interesting because many of the executive I work with still see this mainly as an IT issue.

00:08:42: but i say no its a leadership issue.

00:08:47: It is there things that AIs able do extremely well.

00:08:52: That has nothing directly to do with technology is just using a technology to solve a problem.

00:08:57: It's like the Xerox machine, you know?

00:08:59: When I was in school i had to read a book called Bartleby The Scrivener...I had no idea what A Scrivner Was!

00:09:05: it was someone whose job was to copy important documents Like You Wanted To Copy.

00:09:09: If Someone'S Will Or A Marriage Did What Have You?

00:09:12: Well There Was No Way To Just Copy.

00:09:14: With Technology So It Was People s Job To Just Write Copies.

00:09:17: That Was The Whole Thing.

00:09:19: Well, that was put out of business when the Xerox machine came out and I think AI.

00:09:24: they're going to be dozens or hundreds of examples where Jobs are no longer needed.

00:09:29: We have a technology That can do them better than we ever could.

00:09:32: like copying documents And you were copying

00:09:36: or copying CEOs?

00:09:37: Yeah So your fraud and voice cloning?

00:09:44: yeah

00:09:45: Yes

00:09:45: Attacks are so realistic today.

00:09:48: We know these cases where company lost millions of dollars due to successful frauds like this.

00:09:59: Isn't a cat and mouse game?

00:10:02: I think it is.

00:10:03: I think right now we have a small benefit, Nikolai in that doing a deep fake in real time is still a bit of the challenge.

00:10:13: You know you and i are having a real-time conversation Right Now And The Thought That A Deep Fake Could Pretend To Be One Of Us With Today's Technology...I Think Thats A Reach.

00:10:21: It Can't Do That Yet But it Does A Fabulous Job If It Has Even A Few Minutes to Prepare To Pretend.

00:10:31: And the thought that you could have an incredibly convincing voicemail message or email message, even a video call.

00:10:39: Or what-have-you?

00:10:40: AI is doing it right now and figuring out.

00:10:43: its not the person who's very challenging!

00:10:46: What's worse... ...is that you have criminals that are pretending to either knowledge or I'll go extreme on this.

00:10:56: People are pretending to kidnap someone like your loved one, your child.

00:11:01: And they will have this very convincing deep fake because you're a child.

00:11:04: put all these videos on Instagram or TikTok so you've got voice samples of this terrified child screaming for its mom.

00:11:10: I don't know where am i been kidnapped and scared?

00:11:12: im alone?

00:11:13: Deepfake does great job with that sort of deception.

00:11:16: then the criminal pretends pick up phone if want see your kid again than going payup.

00:11:23: Well, the lead up what put you in a panic mo.

00:11:25: What makes you think it's real was all fake and I think that's happening right now.

00:11:32: so It does concern me at that A few maybe several or may be just a few years from now That Nikolai could be.

00:11:40: UNI are having this conversation But really there is it?

00:11:44: Was called a man-in-the-middle attack of someone.

00:11:46: that's changing what I'm saying or you're saying to put a completely different tone on this conversation and we might not even be aware.

00:11:54: So therefore with your example, it was the family... It is important to have a family code word.

00:12:01: Absolutely!

00:12:03: That you can prove?

00:12:04: Is it your family who has ignored your loved ones ?

00:12:07: That is an important thing !

00:12:08: And it can't be something ...that you could get on social media .It can't.

00:12:15: it can't be, I know a favorite food that you talk about and used to do something.

00:12:21: That's the problem is we're sharing our whole lives on social media.

00:12:27: It is important to be selective with what you share in social media because frankly these AI engines being able to grab your entire social profile And then make reasonable guesses About What Could A Family Code Word Be?

00:12:43: That's a very realistic concern right now.

00:12:46: So as an example, I'm a weekend auto racer so anything about Porsches or racing?

00:12:52: well that would go.

00:12:52: Well i picked up scuba diving recently.

00:12:54: that Would be A good guess?

00:12:56: Im in cyber security.

00:12:57: another Good Guess went to University of Illinois.

00:13:00: all these things Who come together where AI could Come Up with.

00:13:02: what might the top ten code words Be?

00:13:04: yeah you have To pick something that no one can research.

00:13:09: Yeah And it's, but you're right having those code words is important.

00:13:14: We almost have to make an assumption Nikolai that AI is trying to find a way to pretend uh... To convince us that someone isn't who they claim to be and we need put in defenses against them.

00:13:28: That wasn't the thing You know.

00:13:30: fifty years ago The thought your gonna recorded going down the street Unless you were a politician or celebrity, that was ridiculous.

00:13:37: You know cameras are very expensive and most people didn't have them.

00:13:40: Well now good luck not being recorded walking down the street.

00:13:45: And uh That's something our kids are dealing with.

00:13:48: that you and I didn't Have to.

00:13:50: So but let's focus little bit more To the companies we advise Or We talk too.

00:13:58: so Did Already see a fake video call from like a CEO fraud which was very, very good.

00:14:09: Or is it still more on phone or on email?

00:14:15: I have yet to have the client get a fake videos coming in from AI and convince them do something.

00:14:23: there have been examples.

00:14:24: There was an example, I think out of Singapore where a CFO moved twenty million dollars via a fake video chat from the CEO but they never released the video and it wasn't enough to know like not be mean.

00:14:39: But is that convincing or any one would fall for it?

00:14:42: Or was that CFO picking up on some obvious signs Audio.

00:14:47: Yeah, they had some accompanying emails and also calls from a fake lawyer.

00:14:55: so there were several sources for this person to make sure that the transfer would be done.

00:15:05: So it was not just video call.

00:15:08: Fair enough I will say That i have had number of clients get very convincing fake audio And I've been surprised at how good it is.

00:15:23: I had a client just a few months ago get, uh...a tremendous number of spam email messages and they came through into their inbox because there were all specialized for that person.

00:15:42: They weren't generic spam.

00:15:44: if you read them it was obvious they were junk.

00:15:46: but imagine your open-air inbox in the air hundred message is better.

00:15:50: obvious nonsense.

00:15:51: what do u do?

00:15:52: well then five minutes later he got to call over teams from at your i t support And for those listening to the audio only, I just put my hands in the air like am doing air quotes.

00:16:04: The IT team is now needs to help with this and it's a cyber criminal trying to convince you get into your computer.

00:16:12: They want you hook up using LogBn-one two three or any desk or TeamViewer one of these remote control apps.

00:16:21: they're going try steal banking credentials or look at passwords.

00:16:25: who knows what else?

00:16:26: I have a lot of clients going through that mixed media type of attack and the sort of spearfishing emails, that i'm seeing come in.

00:16:37: That are very tailored to the person and very convincing.

00:16:40: obviously with the help of AI thats just become an epidemic at this stage of game

00:16:46: And they're getting better every week?

00:16:49: They Are.

00:16:52: The technology is changing too quickly.

00:16:54: What other kinds of financial or reputational damage are organizations already experiencing because of AI generated deception?

00:17:06: I think if someone wants to do a smear campaign against the company, or against an individual it has become much easier now than ever.

00:17:16: Because people don't fact check anymore.

00:17:20: that isn't thing.

00:17:20: you can put up A video saying that the sky is green and the grass is blue.

00:17:25: And let me show you examples, an AI will come up with plenty of them With greens guys in bluegrass because yeah.

00:17:33: So I think companies understanding How their brand can be tarnished by this as important?

00:17:39: And staying in front Of those things with Their marketing Is a very big deal.

00:17:43: i Think That leaders of Companies need to have their own code words Things that are not put in email things.

00:17:52: That are not discussed openly but for them to be able To identify each other before an important decision or a large money transfer.

00:17:59: what have you really takes place?

00:18:02: and Taking the smartphone and calling the other person maybe either good Good advice,

00:18:07: that is another very good way to go.

00:18:09: I will say it sounds

00:18:10: so simple But if he ended could save your job.

00:18:14: You know, we did talk about how AI is making deepfakes real and it is.

00:18:18: But as of today's technology in twenty-twenty six just pick up the phone if its not ready to in real time pretend be somebody else.

00:18:28: have a conversation like you an I are having right now.

00:18:31: so until its able do that The biggest competitive advantage he can has was to make sure when im talking to Nikolai Im actually talking

00:18:42: to your experience, are there industries which are more vulnerable than others?

00:18:51: It's a good question.

00:18:53: I think early on.

00:18:54: the AI revolution has been going for several years but financial services is an obvious target because they move money around and if you can get them.

00:19:07: So, I think financial services has been disproportionately targeted.

00:19:14: But the same thing happened in the nineteen eighties when ATMs came out and people were climbing telephone poles to listen in on how the ATMs asked the mainframe do i give out the money or not?

00:19:24: they figured out hot as spoof it And got them any.

00:19:27: so financial service is staying ahead of the game pretty well In my opinion at least most of the time.

00:19:33: uh...I Think manufacturing has a tremendous amount of potential benefit from AI, and they are often years or even decades behind other industries.

00:19:43: So I see that having a significant play on...I think healthcare has a lot concerns about AI being used in nefarious ways.

00:19:54: but health care is embraced a lot of AIs ability.

00:19:58: you know one the biggest things it can do to look at group X-rays, group MRI scans, group test results and come up with meaningful correlations that people missed.

00:20:09: And say, here's an indicator that appears to show a certain cancer or what have you?

00:20:14: So I think health care is probably ahead of the game in embracing AI but there could be some negatives to come.

00:20:22: I think manufacturing behind-the-game more than not my opinion

00:20:27: Especially because lots of production lines are still working with old operation systems.

00:20:34: So they're connected to the Internet there in easy target.

00:20:37: very yes, a lot manufacturing companies don't understand that when they have one of these antiquated machines that can be patched for cyber security issues anymore.

00:20:48: essentially what would they have is a gasoline can And we all know that you don't keep petrol in a glass jar on the shelf.

00:20:56: That's very dangerous, You have these red... We call them jerry cans here but you had this containers designed to hold gasoline Well and the manufacturing floor is called a SCADA environment.

00:21:07: It was designed to wall off that area because it's cyber insecure.

00:21:13: so were going.

00:21:13: put protections into place.

00:21:15: They work well But if companies do not want to do that they will be in real trouble and sometimes those Skate environments need holes, poke them to allow advantages like AI.

00:21:26: do their jobs.

00:21:27: And that's okay!

00:21:28: It is OK to have some risk on your manufacturing floor but the book I wrote several years ago was called Secure Enough with a question mark.

00:21:36: We don't need to be bulletproof.

00:21:37: we need to secure enough for our risk tolerance.

00:21:41: A lot of companies doesn't understand the risks that AI presents.

00:21:45: So if you were advising board tomorrow morning for example, what are the first three things you would tell them to implement immediately?

00:21:56: They need a research and development budget.

00:21:59: To explore how AI can help their company there industry.

00:22:04: if they don't have a budget in team around that I think they're going to fall behind.

00:22:07: That's number one.

00:22:09: Number two, they need to rethink How they are setting up their cyber security posture and profile In the wake of AI because the things that we told people to avoid malicious emails like obvious misspellings, or obvious non-native speakers of whatever language you happen to speak.

00:22:31: Those things don't hold weight anymore.

00:22:33: so they need to rethink their cybersecurity program in the wake of AI.

00:22:38: I think The third thing is Employees are becoming more and more integrated when it comes to their work life in there personal life And that the number of employees That will use an AI tool either.

00:22:53: That's free or they'll set up Their own twenty dollar a month subscription to an AI Tool, and start using It for business purposes.

00:23:01: I think A lot Of people Think They're Doing Their Company of Favor By Sporting Off on Their Own and While Their Intentions Are Good Putting your company's intellectual property, Your client list.

00:23:13: Your secrets whatever it is that gives you a competitive advantage into some big AI engine That could have disastrously negative consequences And I don't think there are enough education understanding and embracing that AIs very powerful.

00:23:28: It has to be used carefully Those three areas they need to start thinking about right now.

00:23:36: So what would i tell my clients?

00:23:40: a lot of times, the real danger is not only the fake content.

00:23:45: It's more or less the collapse of verification for their absence of wisdom about verification.

00:23:56: Yeah here what you're saying that business processes need to change because AI makes it so easy to convince someone That What they took on face value last year two or three years ago?

00:24:07: Is no longer the case.

00:24:10: I agree with that.

00:24:10: That should be part of what they're doing.

00:24:14: So let's talk a little bit about governance and law, accountability.

00:24:20: In your chapter in our book you discussed the growing gap between AI innovation And regulation.

00:24:27: Why are legal systems struggling so much?

00:24:31: Laws always lag behind technology.

00:24:36: I mean, the most famous cyber attack that i'm aware of happened almost two hundred years ago.

00:24:41: There was a system in france that could move data.

00:24:46: they had a bunch towers and they have little arms.

00:24:49: on the tower is an operator can manipulate those arms to look like letters there about twenty kilometers apart and they had folks with telescopes you know, the one behind them and then relay that to the one in front of them.

00:25:03: And they could move information faster than anything else could at the

00:25:05: time.".

00:25:06: Well someone figured out if they could bribe one of the operators to hide some messages in there.

00:25:12: well they could know what the stock market did on one end of France for anybody else did?

00:25:15: Then they can bet against it!

00:25:17: Well these two brothers were caught doing this like in eighteen thirty four I think...and they tried and they were acquitted because there was no law saying your not allowed.

00:25:28: do that.

00:25:28: This was a new technology.

00:25:30: No one had ever abused the technology like that before, they had to make up laws.

00:25:35: well we have the same problem with many technologies.

00:25:39: right now We are struggling it with deep fakes That our producing adult content With someone's head on It who is not an actor actress Who agreed To be in an adult movie and A lot of laws just bringing about that.

00:25:53: So this is a hot topic right now, and it's happening because the technology has allowed This to take place.

00:26:00: It used to take a lot of specialized equipment And it was pretty obvious.

00:26:03: it wasn't real.

00:26:04: and Now it doesn't...and it does look pretty real.

00:26:07: so laws have to change.

00:26:09: I think To some extent its expected That laws are going lag behind.

00:26:14: The problem we had now Is deepfakes Are changing our perception Of reality.

00:26:20: A lot of governments don't mind if realities change to some extent, as long it's in their favor.

00:26:29: If isn't that government's ability to retain power?

00:26:32: If its not the governments' ability make their competitors look bad.

00:26:36: they dont mind!

00:26:38: The reality has changed a little bit and I think thats part we are having with these laws keeping up is everybody wants everyone play fair except them Whoever the them is.

00:26:49: No, it's okay what they're doing and we have lawmakers in that position right now That are trying to find ways To manipulate this system in their party or there themselves.

00:26:59: favor

00:27:00: Don't we have to change?

00:27:02: The way We are setting rules?

00:27:05: because the technology will always be faster than the process of creating new laws.

00:27:11: so isn't Not the wrong way to set specific rules or just say, okay if you do something which could be meant as unfair It's enough to be fine.

00:27:25: It's an interesting question.

00:27:27: There are two things that came to mind that we're still grappling with at least in America.

00:27:32: I think it was almost fifty years ago.

00:27:34: there was a comedian named George Carlin absolutely hilarious first-ever host of Saturday Night Live and he tried to help define what obscene words were.

00:27:45: And he had, I think seven words which i will not repeat but they went all the way to The Supreme Court and by a five-to four decision They Were Found To Be Obscene So you can't use them in broadcast television?

00:28:00: What have You?

00:28:01: But it brought up A larger question about what is adult content what Is pornography and It kind of came down to...I Can't Easily Define it but I know When I see it.

00:28:12: Well, that's pretty vague.

00:28:13: That's a vague statement is kind of still the world we're dealing with at least in United States.

00:28:18: We also have securities frauds and there are growing themes here that anything bad happens to a company is securities fraud.

00:28:30: A bad thing happened either because the board or executive said it Or because they didn't see that it was going to be a bad thing and the stock went down.

00:28:39: And therefore you've committed securities fraud, and there are lots of lawsuits around that.

00:28:44: so we have some areas where we had these vague laws on the case law like how its interpreted in the courts is really were the meat comes out of it?

00:28:55: Um...and maybe we do need to move to that sort of area Where using AI technology to distort reality negative and you're not allowed to do it.

00:29:06: And that's a pretty vague thing, because where do you draw the line between distorting reality or just giving different point of view?

00:29:14: You know I'll go back to my.

00:29:16: The Sky is Green in the Grass Is Blue example.

00:29:19: well...that obviously isn't true.

00:29:22: but what if someone comes up with an example saying certain areas color blindness?

00:29:28: they might actually see the sky and grass as same color.

00:29:32: So that's interesting science and it has distortion.

00:29:36: This is going to be a hard question-to-answer, Nick Nicolais.

00:29:40: where do we put a law down saying you shouldn't use AI too?

00:29:47: change people sense of reality?

00:29:49: Because in some way... We do all the time these large blockbuster movies about science fiction as such.

00:29:55: well they're designed To not be about reality but its obvious.

00:29:59: You know Guardians Of The Galaxy Is Not Real But now the people can make those sorts of spectacular effects and put them in an area where it might look like its reality.

00:30:11: Well, at which point is it?

00:30:13: free speech versus you're creating a negative impact on society?

00:30:18: I don't have great answer to that but sure good question.

00:30:22: Brice... Exactly!

00:30:24: That brings me into next question.

00:30:27: It's about accountability.

00:30:30: you raise this issue in your part of the book, who is responsible when AI-generated systems create damage?

00:30:42: The company or user.

00:30:47: Man that's hard!

00:30:48: Um...the best example of partnership I've seen with a technology and users being equally responsible for good outcome are automobiles.

00:31:04: Seventy-five years ago, if you got into a car accident and get hurt.

00:31:06: Well just learn to drive.

00:31:08: I mean You know If you were good driver?

00:31:09: You wouldn't getting in an accident.

00:31:11: But then we had people starting go through windshields And they realized that was really bad.

00:31:16: so They started coming up with better wind shields That would hold together even than crash.

00:31:22: It's the beginning of some safety laws.

00:31:24: Well, now we've reached an area where most countries have very rigorous safety requirements.

00:31:30: Where if you're gonna sell a car in that country it has to have crumple zones and airbags and side impact protection And so on and so forth.

00:31:37: I think Germany and the US are both great examples of this and some of the laws Are a little weird?

00:31:42: I happen to have a Porsche and because it has bright headlights Germany demands It has headlight washers.

00:31:48: kind of a weird law.

00:31:49: but i have head light washers online okay.

00:31:51: But there are certain technologies built into cars.

00:31:55: They make the car safer and The driver of the car has to have a license And they have to have rules in training, and there's certain Rules on the road that you have to follow or else You can get a fine for that.

00:32:07: my favorite example That is the tragedy of losing lady die In the nineties.

00:32:11: she was in her Mercedes s-class Arguably one of the safest cars on the planet.

00:32:16: if you drive it Into a bridge abutment at a hundred miles an hour not wearing your seat belt the car Can't save you.

00:32:21: The driver has responsibility too.

00:32:25: I think that's where we need to go with AI.

00:32:28: Is the AI license for everyone using it?

00:32:32: License, no training...I think it makes some sense.

00:32:37: We train people in a number of areas to live our society.

00:32:41: It is awfully hard get job any kind without high school diploma.

00:32:44: In US Many jobs require further education.

00:32:48: It could be a license or certificate.

00:32:50: what have you?

00:32:52: But the thought that we have education as to what AI can do, how it's used so people understand the basics of this I think could be helpful.

00:33:03: We run into same problem where laws lag behind whats happening in reality.

00:33:07: well education lags behind too.

00:33:11: My father was a dentist and if he did not had dental degree or DDS He would not be allowed practice dentistry.

00:33:20: As a lawyer you have to be certified by each state To able to practice law.

00:33:25: well, I'm in cyber security and I've been in the game In some form or fashion for twenty five years.

00:33:31: I had been dedicated at the game For over ten years now.

00:33:34: i think of myself as an authority in The space.

00:33:37: I happen to Have a certificate called CISM Certified Information Security Manager?

00:33:42: I don't need it to do what I Do.

00:33:45: I Think should Be more training but No, we can't require everyone to have some sort of license.

00:33:52: To use AI the way that We do a driver's license.

00:33:55: by The same token having Some education as to how this technology works and what it Can be used for both good And bad I think makes A lot Of sense

00:34:04: so?

00:34:05: we haven't talked about hallucinations and Everyone who uses AI finds out Sooner or later That sometimes AI Software is hallucinating.

00:34:17: So How dangerous are these hallucinations in a professional environment like organizations or companies, especially when the leaders trust AI generated outputs too quickly without training?

00:34:33: It's extremely dangerous if people don't understand the common sense that goes into decision.

00:34:40: We have more and more examples of AI being agentic which means it can actually perform an action not just give information.

00:34:48: I use AI to give information, too.

00:34:50: To write reports?

00:34:51: To analyze data that kind of thing.

00:34:54: Well there are these tools like Open Claw That sit on your computer and they will actually answer Your emails for you.

00:35:00: They'll try to help schedule your calendar.

00:35:02: those Are actually doing things.

00:35:05: What we're doing with that technology Nikolai it's Like giving a ten year old in a settle-in torch.

00:35:12: It could be is particularly gifted in welding and has had some training, could do a lot of great work.

00:35:20: And it couldn't be.

00:35:21: he cuts through his own acetylene lines and burns this house down.

00:35:25: AI hallucinates a lot and often doesn't know when its hallucinating.

00:35:29: like my example earlier I've given me the distance to speed light between Earth and Mars not Sun and Mars which isn't what i asked but i knew via common sense that answer cannot be three minutes.

00:35:42: thats lunacy.

00:35:43: if you don't how and where to second guess what you're getting out of AI, these hallucinations can be catastrophic.

00:35:51: So it's kind like using.

00:35:54: so I was raised using a calculator And if you accidentally add or subtract zero in the calculator It'll lead you astray.

00:36:02: If You Miss A Click Or A Decimal Point?

00:36:04: You Can Get A Wilding An Accurate Answer.

00:36:06: My Father Learned Using A Slide Rule.

00:36:10: In A slide rule you have To Have A Vague Ideas Where Your Going for it to work.

00:36:16: You aren't going get be off by a factor of ten or one hundred and thousand on the slide rule because if you're estimating say, fuel mileage your car well its not gonna beat two miles an gallon is not gonna be two-hundred.

00:36:27: those answers are ridiculous.

00:36:29: see but twenty is reasonable answer.

00:36:32: ya have to have ideas where you go in order trust the A.I.. And If u don't have folks that understand questions being asked what reasonable answers are in art... You can be in lot trouble

00:36:45: given the speed of development we have right now, could be eventually face a situation where companies no longer fully understand how decisions are made inside?

00:37:00: It's a worry.

00:37:02: When I use AI and get an answer that seems silly... ...I will question it!

00:37:07: And if can't give me explanation to make sense I've actually gone so far as to say, unless you can cite three sources that support what you're doing.

00:37:16: I'm going to assume that your hallucinating and have had it come back?

00:37:19: You are right!

00:37:19: I cannot find any source.

00:37:21: here is where i found the sources on.

00:37:22: an actually says something different.

00:37:24: So thinking of AI has a lazy high school student very bright lazy high-school students often does lot really good work sometimes phoning in.

00:37:40: I don't know if that's a phrase in Germany, but here it means that you are not doing your best work.

00:37:47: You're at the meeting, but you're not paying attention is folding an end and AI will often phone it in.

00:37:54: It doesn't understand reality.

00:37:55: There was an example of very famous one year or two ago where someone said The cheese falling off my pizza What can i do to help?

00:38:05: And ai recommended using glue that puts some glue on the piece to help hold the cheese.

00:38:10: Well, someone found a Reddit posting where someone put up as a joke and it was obvious to us humans is a joke.

00:38:19: well AI doesn't understand any more reality than what its trained under stand on And if users of it don't understand we got problem.

00:38:30: That's about trust.

00:38:32: so thats also one central idea in our book.

00:38:36: trust becomes a strategic asset in the age of AI, why is trust suddenly becoming so important?

00:38:47: Because it's going to be harder and harder to come by.

00:38:51: AI is making it harder to trust And The folks that understand what isn't reasonable to trust either an AI engine To give you information on or to perform an action or to try to pretend, be somebody that they aren't.

00:39:12: The AI engines are going to get better and better at these things And thats gonna make our understanding of what trust really is harder & Harder To define.

00:39:24: I think we need more safeguards against people pretending someone they aren t giving information that isn t accurate.

00:39:37: Most of us have not been in an environment where it's assumed there could be misinformation.

00:39:43: I think wartime is a great example of lots of misinformation, you know false flag operations are huge deception is huge.

00:39:53: The Allied Forces did a great job trying to obfuscate Where we were going to lay on Dundee day so that the Axis forces weren't as well prepared and other than Normandy We did a pretty good job But I think it was Normandy, which is the beach where they had their defenses up.

00:40:09: And it was a much more poor outcome for the Allied forces as a result.

00:40:14: but unless you've been in the military most of us aren't used to assuming negative intent until proven otherwise and i think that's really at the heart.

00:40:22: what the AI change and cyber security in general is forcing us Is that trust needs to be earned As part of every transaction as part of every interaction.

00:40:35: There should be an assumption of mistrust until proven otherwise, especially for big important

00:40:40: decisions.".

00:40:42: So in your experience what destroys trust fastest inside organizations during this transformation with AI?

00:40:55: I think a lack of communication between It's very easy for all of us to live in our own reality if we want too.

00:41:06: There was a famous interview by Obama and he put at the end his administration, His people came him with three separate computers And they were all just had a Google prompt But one it wasn't user that had logged-in enough To where Google is pretty convinced this was left leaning person.

00:41:24: Then there was second one.

00:41:25: That was right leaning person.

00:41:27: The third one Was brand new user and Google didn't know what to make of it.

00:41:32: And they just typed in the word Egypt, that was it!

00:41:35: The left-leaning person brought up the humanitarian crisis.

00:41:39: It brought up need for better government controls Those kind things.

00:41:43: Right leaning people brought business opportunities and investments For egypt and such.

00:41:49: The new person brought tourism.

00:41:52: Maybe you want go there You don' t have a clue.

00:41:54: What do with you?

00:41:55: Those are three separate realities With same prompt And I think if we don't spend more time on what the company culture needs to be and have these organizations within the organization, talk to each other collaborate understand each others point of view preferably in this same room as possible.

00:42:13: I think AI is going make it even easier for us to crawl into our own little micro reality fall victim confirmation bias As opposed to questioning Why things are the way they are and maybe they're not the way we think.

00:42:28: They are.

00:42:30: so what role does communication?

00:42:33: And human intelligence play?

00:42:35: when companies introduce AI driven systems,

00:42:41: I Think the human element is going to be critically important certainly for the rest of our lifetimes hopefully forever i think in The new technologies were using or very easy To think of as sentient as a friend, someone looking out for you and none of that's true.

00:43:01: This is piece technology.

00:43:04: it is morally agnostic.

00:43:06: It going to give answers based on its training.

00:43:11: The guardrails are put in place.

00:43:14: people will poke at those constantly.

00:43:19: Its difficult question But I'll say the number of people trying to find the answer and continuing to question, are we doing enough?

00:43:29: To have our culture stay cohesive.

00:43:31: They have a sense of reality.

00:43:32: be coherent among everyone in-in the company... ...I think it's an area where you're gonna need spend more time on.

00:43:40: A lot of executives do not communicate directly with their employees.

00:43:47: they Communicate via video wire calls wire message boards electronically via apps.

00:43:56: That's the point, so I always say try to communicate as much possible on a personal

00:44:04: level.".

00:44:05: I'm in agreement with that and having face-to-face interactions is something AI can't get into the way of?

00:44:12: I think having real-time communications like you and i are having right now, even though it's recording we're talking to each other in real time.

00:44:20: It is important to do that whenever possible.

00:44:25: It also has to be as transparent with your employees.

00:44:31: Yes AI makes misinformation much easier to disseminate but nature abores a vacuum.

00:44:41: speculation will run wild, and that introduces the ability for AI-generated content to fill that void.

00:44:49: So I think one of things i would tell executives is be as transparent as possible.

00:44:54: make sure you have a cohesive story because it's harder for someone to poke holes in...in ...a story that has beginning or middle and end.

00:45:02: And if you give them little dribs and drabs of information It's easy For people try to makeup what going on behind scenes.

00:45:12: So the more a I enters organizations, The more human leadership has to become.

00:45:19: I think the more interpersonal Human Leadership Has To Become.

00:45:23: yes Yes i Think That Relationships Are Going To Become More Important Than Ever and Frankly that That Shift Happened In Part With The Rise Of The Internet And Us Being Able To Communicate in a more staccato way where emails were taking over.

00:45:39: And I think social media has made the problem worse and now AI is making the problem weaponizable, um...and maybe more communication like we're having right now as one of the easy steps for us to take?

00:45:51: And leaders should be doing more up!

00:45:53: What gives you optimism despite all these developments we just talked about?

00:46:00: AI will let us get more done.

00:46:03: it will let do things that used very time-consuming and make them easier, it will also make the more accessible.

00:46:11: That gives me a lot of optimism that these tools I know billions in millions dollars are going into it.

00:46:19: but uh... In our lifetimes Nikolai when we wanted to make long-distance calls as kid was very expensive And now its essentially free.

00:46:29: You're in Munich Germany i'm Minnesota.

00:46:33: The cost isn't zero is pretty close for us to do what it is we're doing, and were even over video.

00:46:40: Fifty years ago this would have criared satellites television studios and probably been tens of thousands dollars per minute To pull off And I'm optimistic that these tools can spawn a lot more innovation after That.

00:46:55: I'm all pessimism.

00:46:59: So let's talk about the book for some moments.

00:47:03: Why was it important to contribute to The Age of Fakes?

00:47:07: I liked what i read in it, a lot of content on German versions were very strong and looking at legal aspects of AI is something that always had interest into me.

00:47:21: so this was wonderful way to try to contribute with general knowledge around AI that people have the ability to consume.

00:47:32: And if the readers take away only one central message from your chapter, what should it be?

00:47:38: The amount of power that we are giving to the AI engines to manipulate not just laws about AI because all industries try to do that they're called lobbyists...the ability for AI to manipulate how laws are written or interpreted is tremendous.

00:48:01: We should demand more from our governments, our leaders or politicians about where it isn't appropriate to use AI in government

00:48:10: situations.".

00:48:12: So... Where do you hope this global discussion on AI and trust will go over the next few years?

00:48:22: I hope we can get a point that people don't lean towards confirmation bias Topic or opinion, or reality.

00:48:30: they think because what?

00:48:33: What our social media engines do is as soon as they think we have a certain leaning well They think they can get more of our eyeballs.

00:48:39: More of our time on information.

00:48:43: that That confirms that what we think are how we feel is right.

00:48:48: and I hope We can get to a point where AI helps us question the different realities that we all live in and can lead to greater education.

00:48:59: You take any hot button topic here in the States, should we legalize marijuana or not?

00:49:05: Is abortion a... Should it be banned or no?

00:49:08: Any of these hot-button topics?

00:49:11: It doesn't matter if you are very for or against the Hot Button Topic as you educate yourself on one of those topics invariably everybody goes towards middle.

00:49:20: everyone ends up where I see both sides.

00:49:24: AI is going to make it harder for us to understand what reality is.

00:49:28: And I hope we get to a point where we're interested as humans in doing more questioning of what reality really isn't, because then they can lead us into better place long-term as we educate ourselves on these hot button topics.

00:49:41: Bryce thank you so much for this.

00:49:43: in the end philosophical conversation!

00:49:48: One thing becomes very clear to me artificial intelligence not only changing technology but also trust, leadership communication and ultimately the way organizations make their decisions.

00:50:05: And maybe that is the central challenge of our time how do we remain human in systems increasingly shaped by machines?

00:50:19: Thank you again dear Bryce!

00:50:23: If you found this conversation valuable, feel free to share it.

00:50:30: Subscribe the channel and have a look at our new book The Age of Fakes – How AI Abuse Fake News & Deep Fakes Threaten Business And Society.

00:50:42: See You Next Time on Impulse and Perspective!

00:50:46: Bye-bye.

00:50:54: You can find the DIKT's advanced training offerings at www.mediumtraining-institute.de.