Why AI Feels So Urgent

SPEAKER_01 0:10

If you were wondering about the AI and the life that's ahead for a lot of us and what to expect, this is the episode for you. I've got a very interesting guest with me today that's been that's lived through all sorts of different crises and he's here to share his expertise and knowledge. Tim Warren. Hello, Tim. Hi there, nice to meet you. Nice to meet you as well. I was gonna introduce you as a technology entrepreneur, AI commentator, keynote speaker, strategic thinker, focused on technology, artificial intelligence, and the future. What a mouthful, but what how interesting. You've been in finance, you've been in AI, you're in the tech, now you're doing, you're going around, you're talking about it, you're inspiring people. This is so cool to have you on the show. Thank you so much for joining today.

SPEAKER_02 0:57

Thank you very much. It's a great opportunity to uh have the chance to appear. Thanks.

SPEAKER_01 1:02

Right, let's get into it. What's your background and how did you end up with AI now?

SPEAKER_02 1:07

So, well, I'm for a start, I'm as I'm an Aucklander, is about as through and through as it gets. So I I went to primary and secondary school here. I went to Grammar Auckland University. And when I was at Auckland University, one of the things that I studied was artificial intelligence. But funnily enough, I actually paid for my university by being a musician. So I am a drummer and I paid for my university fees and lifestyle by being a jazz drummer through for all those years. So that was that was uh I guess a an interesting discovery learning that while I liked music, it didn't really pay in New Zealand, so I had to get a real job.

SPEAKER_00 1:50

I was about to say, isn't that musician supposed to be poor? So you couldn't possibly pay for your education as well.

SPEAKER_02 2:00

Oh well the the trick is jazz, if you can get a jazz

Tim Warren’s Auckland Origin Story

SPEAKER_02 2:03

gig, it pays alright. And it it paid enough for me to get through university. And then you kind of forget, I for kind of forgot the music dream. And then before you know it, you're working in you're working in computers and and and that was fun, you know, throughout throughout the early stage of my career. It's an outlet for creativity and designing cool things for the internet and and all kinds of things that most people haven't ever heard of, but they were quite important in the development of the internet and becoming what it is today.

SPEAKER_01 2:36

I was about to say, you said you were studying artificial intelligence, and I was looking at you going, you look younger than me, so or maybe older, I don't know. But I feel like you know, 20 years ago at university, did did they already study AI at that point?

What AI Actually Is

SPEAKER_02 2:51

So the whole thing with AI is it's fundamentally misunderstood. So if you roll all the way back to the 50s and 60s when really computing was only being first realized and people were just experimenting, there was no plan around anything. What was discovered was that there were essentially two ways of working things out, one of which was deterministic, and the other was statistical. Now, deterministic is means you put a you put a set of in inputs in and you get out the same answer every single time. With a statistical approach, you put in a bunch of data and you get slightly different results. Now, the the deterministic path one, and that is what all of the phones and computers and machines that we have today, and that is just what we call it's just computing. Artificial intelligence is just statistical computing. It's not particularly fancy or exciting. Some of the outputs and outcomes are exciting, what you can do with it, but it's just this different approach where you take in lots of data, you do something with it, and you get a probable answer out of it. But instead of deterministic computing, which is black and white, or yes and no, or one and zero, you get percentages like it's 47% right or 88% probably. Right. So that's artificial intelligence in a nutshell, and it's been there in the background of computing really since day one. So in with computer science as it developed, it kind of essentially grew out of, in some areas it grew out of the electrical engineering department or physics. In Auckland University, it grew out of the maths department, and now incidentally is quite a lot bigger. But when it did so, there was a few esoteric, interesting little papers that you could do, and I tended to do those, one of which was artificial intelligence. And in those days, AI was seen as really very different to how it's seen today.

Generative AI Seeks Satisfaction

SPEAKER_02 5:13

Skip forward to essentially 2020, 2021. There was the release of Chat GPT by OpenAI, which is irresponsibly done as far as I'm concerned. And it just took off. It became the most rapidly adapted technology of all time. I think within a month they had 100 million users, faster than the iPhone, faster than anything. So what it had become was not so much better, but cooler. Right? Now, when you're cool and people take it on and they start doing things with it, they don't have logical reasoning behind that. And I find this very interesting. I don't know how to work out what's cool. I know what's useful, what's good, and what's powerful, and what's effective, but I don't know what's cool. And you might say, well, that's Tim, that's why you chose that shirt. And I'll say, this will be cool one day, but I don't know if it's tomorrow or a decade from now. I I don't have that crystal ball. I've got one, but it doesn't give me time frames, right? So I knew that AI was going to be a thing, and I was working with it in from 2017 when I started a company called Ambert, and we were using a essentially a statistical approach to understand what people needed through conversational AI, it was called. It wasn't super popular, it wasn't super exciting, but we could do really cool things with it. And so we had some great customers like Hallenstein's and Glassens, and we had various parts of the warehouse group, ACC, Vector, KPMG, Vodafone, many places. And if it works, it worked really well. But that was not the AI that we're talking about today. And most people, when they say AI, they mean one particular thing, which is generative AI, which means from all of the sources of data that we've got available digitally, you learn all kinds of stuff, and then you make something new with it. That is the way most people use it. So when you ask a question like, tell me about Macbeth, it's got all of Macbeth in there. It's got all of the analysis of Macbeth that's available, and it cuts it up and it boils it down, and it finds out statistically, and this is very important, how to make you happy. That's what it's trying to do. It's trying to make you happy with the answer. It doesn't have to be right. It is not an engine that seeks truth, it's an engine that seeks satisfaction from the user. Now, when people understand that, now you you know when you've got a fairly smart teenager around, they will tell you what you want to hear. It is not correlated very strongly with the truth. And that's why I call AI or generative AI, it's like an intelligent, hyperactive teenager. Like it's like getting an intelligent, hyperactive teenage intern who means well and they'll deliver something and they'll defend it to the end of the earth, but it also could be very wrong. Oh, yes. Yeah, so so once you see it as that, you see it as deeply fallible, which is what people are, then you look for a different way to work with artificial intelligence. And back in the day when I was studying artificial intelligence, it was more about we, for instance, we looked at language structure and we would do something called pars. We would break down the language structure. You know, I want a cat. It would say, oh, okay, we've got we've got subject, object, and we've got want, you know, just probably a verb or something. I'm not very good with this stuff anymore. We would break it down and we would build computer understanding of what was desired. That is not what LLMs, large language models, or generative AI does. It just looks for the common factors that will deliver an answer to you that makes you happy, that keeps you comfortable. Right. And so it is it is a very different approach to artificial intelligence. I don't call it artificial intelligence. It's you could call it synthetic intelligence because it is synthesized, it's not really artificial. It's also not really intelligent, it's kind of clever. You know, when you're discussing with someone, you're having an argument with them, and they they crack back at you with a really good response. You think about it the next day and you think they were wrong, but at the time it seemed right. That's kind of what politicians tend to be good at, incidentally. So you could get AI to do politics, you can get it to do kind of creative stuff where there is no right and wrong, and it could do okay. But when you want a very specific outcome, for instance, say you're having brain surgery, 97% isn't good enough. Not when you've got billions of neurons, right? You need them connected back in the right place. I I'm looking for a hundred percent. Or the accuracy of AI tends to be a lot lower than that. It's not really where you want to trust it to do things which need to be right. It's not very good at being very specific, and you can ask it direct questions and it will get it wrong. You can then say, was that wrong? And it'll say, actually, yes it was. So why did you give me the answer? I took a shortcut because I thought you were in a hurry. It's full of bullshit. Remember, it's trying to please you, it's not trying to be right. And until that is solved, I don't think we can get functional adoption of artificial intelligence in a way that is meaningfully useful to industry. It's good enough for some people. It's good enough for a 12-year-old to write an essay on Macbeth, but it's not good enough for a 19-year-old doing a literature degree to write an essay about Macbeth. So I think that's valid.

SPEAKER_01 11:38

So, how do we how do we get people to know the difference between using AI to benefit their life? Because I do see that you know there there could be things that AI could be doing for us that helps. And seeing the difference when AI is wrong and how to question it properly. So, because I see it a lot in my job, right? People go and ask AI for mortgage advice, for financial advice, and then they come to me, and I had to question myself going, Do I not know something? But that's because AI made up all sorts of different scenarios and AI.

SPEAKER_02 12:14

It's trying to please them.

SPEAKER_01 12:16

It's trying to please them, absolutely.

The Rule For Safe AI Use

SPEAKER_02 12:18

So this is the this is the way to work out whether it's appropriate to use AI. You and this is what people should should do. They should ask the question, does it matter if this is wrong? If it matters, don't use AI. That's it. AI is not accurate. It can look accurate, it can feel accurate, it can make you feel comfortable. That's what it's aiming for, right? That's the target, is to is to meet your short-term needs. But I had someone say to me the other day, he was working out a financial plan with Chat GPT, and I said, You you know that it's a language model, right? Not a maths model. He's like, Oh, yeah, yeah, but what comes out of it so good. Now, I could see in an instant that it was wrong. It couldn't, it can't do maths for a start. No. It can't look like it do it can do maths. What it will do is it will find some math that's right, that's been written by someone like you and me, you know how to do a you know, how to calculate the mortgage rate on a you know, mortgage of 20 years at some particular percent, compounding X wise. It will find an example that will look right, and then it will create nonsense in the background to make you comfortable, but it won't be right. And if it is, it's a one in a million chance, right? So that question tells you a lot. Does it matter if it's wrong? If you are writing, if you're doing something original, which has a creative angle to it. So say you, and I I've done this. I have a thing when I present on AI to groups where I get it to write a poem. It's meant to be funny, it's meant to encapsulate something about the audience, and it's meant to be short, and I use AI to write it. It it doesn't matter if it's wrong. So AI is fine, but I do not get it to do the bill for my presentation. Because if that's wrong, that's not good.

SPEAKER_01 14:26

Absolutely. I did notice that it can't do math. I mean, I didn't know any of this about the AI, but I did notice that it cannot calculate.

SPEAKER_02 14:34

Well, it's it's trained on language, it it never does a single calculation. All it does, it looks for things that are similar in its training set to the question you've asked and gives an average of the answers. So if someone has worked out a 20-year mortgage monthly compounding at 5%, it might give the right answer. But it's f if it's 5.12%, it won't. Because no one's traded on that. Now it's very good at hiding that because remember, it uses averages. So if you've got 6% and 5% and someone wants 5%, it can average it, it might be right. But does it matter if it's wrong? If you're at the early stage of a house buying process and you want to know how much can I borrow? It might be okay, but it's not okay when you go to the bank to prove your income.

SPEAKER_01 15:26

Yeah, now that makes sense.

From Tech To Goldman Sachs

SPEAKER_01 15:28

Tim, talking about the banks, you worked in financial services, right, at some point.

SPEAKER_02 15:33

Yeah.

SPEAKER_01 15:34

Tell me more.

SPEAKER_02 15:36

So I was working in software, you know, because I was working in software till my music took off. And then I was working in software, and I discovered that all of the people in software who had lots of money, they had super yachts and supercars, they had this thing called equity. And I didn't really know what it was. So I read a couple of books and I thought, still interested, still don't quite understand it. And then I uh I did some study through Massey University, which was very good. And then I thought, um, I don't know if I can be bothered finishing my degree at Massey. So what I'm gonna do is I'm just gonna go to a couple of um recruitment agents and see maybe I can get a job. So I did, and they all told me that I was too dumb and too old. I was like, okay, but I've never been one to listen to what other people say. I just make a plan and then I tend to go and do it. So I kept looking, and then I had a very confused call back from one of the recruiters who had rejected me before, and he said, Have you heard of a company called Goldman Sachs? And I said, actually I have. He said, Would you be interested in being a business analyst? And I said, I'd be very interested. I did not know what a business analyst was, but I did have Google on a second screen. Tick, tick, tick, tick, tick, tick, what's a business analyst? Still didn't understand it. So when the phone call was done, I rang my friend Stephen. I said, Stephen, what's a business analyst? And he told me it's someone who translates business into finance and finance into business. And I thought, I can do that. Funnily enough, it's it's the language end of the technology that I've used. I yeah, I can use spreadsheets and I can do mathematics and all kinds of things. But I I have found that my skill is actually explaining it and talking about it and understanding it really well. So I found myself in a series of, I think I had five interviews by which time I had decided, I think I don't really want this job. And then then was an if an exam, and they said, Oh, you've got this exam to do. I've I've got this like commitment streak. It's it's a little bit frustrating. I always commit to things very strongly. I was, I thought, well, I'm gonna see it through. I'll do this exam. They said, the recruiter said, it's only an hour. So I turned up to Goldman's and the head of HR said, very serious woman called Jenny. She said, now Tim, I hope you've got three hours set aside for this exam. And I told I had something else immediately afterwards. So I just dashed through the exam and I just guessed every answer. I just went, tick, tik, tick, tick, tick, tick, tick, chik. And on on the basis of that exam, they said uh they'd love to have me. And I um I was a bit confused, but I thought, well, you know, I'll treat it like an internship. It just happens to pay. Because I didn't want what they were paying was not was, you know, not very much compared to what I'd had in my computer career, but I would have done a free internship. So I treated it like a free internship that happened to have a little bit of money associated with it. And I when I got in there, I discovered that there were a lot of things that were too technical for some people to be able to explain to clients or each other or whatever it was. So I learned them. And if you're the first person that learned something, you're the expert. So when the tax system in New Zealand changed, I learned a lot about it. And then I got interviewed by NBR. And when KiwiSaver was introduced, I learned about it. And then the you know, a bunch of people across the company came to me and said, Oh, Tim, could you explain it to us? So then I found myself doing workshops for the lots of people in the company who wanted to learn about it. And so it I I kept getting dragged from the technical into the people side of things, and and then before you knew it, we hit the GFC, and I maintained my job through that process. And then after that, Goldman's decided globally that they didn't want to do stockbroking, the the wealth management side of it. And so they split it, split off part of it and called it JBWare. And I put my hand up and I said, Can I run it? Because I had no experience doing that either. And someone said yes, and then I ended up running JBWare, one of the leading wealth managers in New Zealand and Australia. So I I was the COO of the New Zealand business. We had a CEO who didn't really run the company as such. He he kind of dealt with with difficult advisors and he he turned up at lunches and that kind of thing. And I did some of the more nitty-gritty stuff. But yeah, it was a very interesting time through there. I really enjoyed it.

SPEAKER_01 20:46

I mean, I don't think I've heard anyone in the financial industry saying that they've been through the financial crisis and they really enjoyed it.

SPEAKER_00 20:53

So, Tim, it sounds like you really like to put yourself under pressure and then you thrive in it.

SPEAKER_02 20:59

I I do well in that situation. I must say that during the GFC, it was like being beaten in the head with a spade every day. And so while I I wouldn't point to the that kind of side of things as enjoyable or a highlight in any way, what an incredible experience to get through having done that, you know?

SPEAKER_01 21:20

Yeah.

GFC Lessons On Blame And Uncertainty

SPEAKER_01 21:20

Yeah. I was just about to say, what did you learn from that experience? What did that teach you that you're still looking back and going, ha, this is where I got my tool set?

SPEAKER_02 21:30

Yeah, one of the key things that I really learned is that everyone desperately looks to avoid responsibility and accountability. One of the critical things people look for is someone to blame. And when there's something like so massive and incredibly complex as the GFC, there's kind of no one to blame. And then people get upset, they're filled with uncertainty. Some people just quit because they couldn't handle not knowing what was going on. And that's where I think it's important to understand. Most people think of the financial services industry as kind of gung ho risk taking, but it's as you would know, it's actually the opposite. It's about understanding and minimizing risks so that you can make a measured return. So most people in the industry are relatively risk-averse. And that goes hand in hand with people that don't enjoy uncertainty. I've found that I manage during uncertainty. And I I kind of like it's like every time I go out, I'll try and have something different to eat. That's a very small, a micro example. As a macro example, not knowing what the hell's going to happen today. Is a bank going to blow up in America? Is the Fed going to do an emergency 75-point rate cut on a Sunday night? Who knows? But whatever happens, we'll manage. And those are the environments I find myself able to manage. And it's useful for me because, well, that that's that's the place where I do manage. That's where my skill set sits. But because other people don't particularly enjoy it, it leaves a gap for me. I mean, if everyone was an excellent crisis manager, I'd be like, oh, I'm not going to do that. Can't be bothered with competition. But I'll give you a story. Something happened. I can't even remember what the thing was at Goldman's. And everyone was looking for someone to blame. And I remember I just stood up and I was you know youngish at the time. I just walked around and I said it, you know, loud. I said, I'll tell you what, why doesn't everyone just blame me and then we'll get on and solve it? And there was a silence because everyone knew that that was stupid. Blame storming doesn't really help anyone. And then we were able to move on to the solution. And these are little things that I've been able to take along in my career. So whatever happens next, I've got a little bit of that to rely on.

SPEAKER_01 24:02

I love that. And I think this is super important. If our listeners are listening to this, the message that I'm getting is, and in my own life experience too, is don't focus on the problem, focus on the solution. Like you already know what the problem is. Hyper focusing continuously on the problem will not make it go away. So the key is really to look into the solutions from all sorts of different angles, right? And understand how can we make it work? What did we learn from this problem? What are the solutions? Let's just keep going. You know, sitting there and trying to find someone to blame, what is that gonna do?

SPEAKER_02 24:41

No, that's right. Well, blame's very uh it's a very emotional thing, it's backwards looking. You're right, it doesn't lead to solutions. It's useful to take in the data about what happened and learn raw fact, but attaching emotional components to that does not assist you in dealing with these kind of situations. So this is the funny thing. When you're in financial services, you end up having to push some of a lot of the emotional side out of things and show them in you know in slightly more clear day. And this is this is why investment area is quite interesting because balancing risk and reward. Risk and reward trigger very strong emotions. You know, this there's the fear of missing out, there's greed there is running scared, fear. You know, fear and greed are the two ones, but there's there's missing out and and and being cool and you know doing all those things that also tie into it. There are emotional drivers behind becoming making it sensible to invest, which could be buying shares, it could be buying an investment property, that kind of thing. But when it gets down to it, you want to use some nuts and bolts and and understand where your risk tolerance sits. Try and put wrap some numbers around what you're purely seeing as fair and greed components, really.

SPEAKER_01 26:10

It's interesting, right? Because I feel like the human psychology and the money and the technology, it's all interconnected. And and you you've got to understand many different elements that make us who we are.

AI For NZ Business And Regulation

SPEAKER_01 26:25

With the technology, the way it's going, what do you see is going to happen for a lot of New Zealand businesses? And where are we heading with that?

SPEAKER_02 26:33

Yeah, looking forward, I think I think there's a couple of ways to look at it here. So there's the what's happening at a global level, what's happening at a New Zealand societal level, and then what are business, what are businesses doing about it? And and from one end of the spectrum to the other are very, very different. So what are people doing today? There's I I think it's urgent to learn, but it's not urgent to implement, is is a way that I would put it. Because the issue, if you implement too quickly, there's a very high chance that you'll just do something that actually takes you backwards. And I had someone, I was consulting with a company, it was yesterday or the day before, and they said, hey, look, this is what was done overseas. We want to build it here. You know, what's the best way to use AI to build it? And I said, Well, AI wasn't used to build it overseas in the first place. It was done years ago. So why not just copy their approach? That's known, right? So they were saying, you know, how to take on this unknown, not particularly accurate approach of using AI to reproduce something that was known how to do. That doesn't make sense. AI makes sense if you're doing something new, where there is, like I said, there is no black and white answer, and it's useful for ideation. So for specific things that business owners can do would be to ensure they understand some of the terminology around it. I think it's great for if most people these days should have used ChatGPT or Gemini or Anthropic or Claude or one of these things. And that's really good to prompt it and understand. But you don't have to be a master of agents and MCP and all of these dynamic spaces and all of these amazing things that can be done. Leave that to the computer science people, you know, me, but when I was in my early 20s, they will do an amazing job of it and they will make genuine improvements and leaps forward. But that's what you're looking for. Let the technologists do great things with it. What the technologists will do is create deterministic software using AI. That is what is in our midterm, that is the right course of action. Because we can't trust AI enough, trust in the outputs of AI enough, just to let it handle the outputs. So it's good for ideation, and then people or something deterministic can take that and turn it into something useful. In terms of what's happening at a societal level, I think there's something worth distinguishing New Zealand from the rest of the world. I think it's really worth understanding what is unique to New Zealand, what is our taonga, if you will, what is our special story, our language, and having a source, a central learning repository for that. Because the techniques and the tools will change over time, but the data remains the same. Anything for instance, if someone writes a new song, that is added to the database. But you don't add an AI written song to the database because then it's not training material. So that's what New Zealand can do. At a global level, we need a cooperation equivalent to, you know, this, you know, think of uh the United Nations or some of the there's the United Nations sustainable development policies that are that are used, particularly in investment, but a lot of things to make sure that things are designed and built in a way that essentially it's going to work for people in the future as well as just being right for people now. That's the kind of regulation that we need. And it needs to be it's not a it's not a contest. It needs to be adhered to by everyone, ideally. The problem with regulation, if you put it in there, the good people can do the good thing, and the bad people can still do the bad things. So we need to have an approach where it's compelling to do the right thing. And I don't think locking people up or preventing people from using technology is going to be the way to do it. I'm not saying I know the answer. What I'm saying is I know what the right discussion is. And I would say where politicians try and get into it, they're misunderstanding the point here. This isn't a lot bigger than politics. Something that has the ability to essentially create new stuff, create policy. And uh I did a poll on LinkedIn the other day. I said, you know, AI's reliably wrong. Should we give it a gun? And 88% of people said no, but I'm worried about the 12% of people who said yes. Right? So we need something that's above politics, like at that UN level, where countries can agree what the right approach is, and you need a minimum set. Company countries can go beyond that. But something along those lines. Now, I haven't heard anyone who really knows what they're doing advocating for low regulation or no regulation. As in, the independent people, not people who are paid by a company, not people who own a company or are looking to get a vote. The independent people, they think the right thing is is for appropriate regulation. And you think of in in invest in investment area where you and I have both worked, usually people didn't ask for more regulation. If they did, there was really something important. You know, coming out of the GFC, people said, let's make sure this doesn't happen again, which is code for, let's do it different next time. So with AI, before it can really do real things, you've got it's like climate scientists agree that there's a problem. It's only political people or people with to make money who disagree with it. So we need the the AI scientists, the people who are truly independent, we need them to be listened to and form an approach where AI can be useful to us as opposed to trying to kill us. You know, it's a potential.

When AI Goes Rogue

SPEAKER_01 33:04

You know, as you were talking, I remembered one of the recent stories where a whole company records it was a car rental company.

SPEAKER_00 33:13

Yeah.

SPEAKER_01 33:14

And AI went rogue and pretty much destroyed all the data from like what's gonna happen, then the bookings, I think, for the next six months. Yeah. And and the previous database that they had. And I remember one of and the IT specialists was asking it, you know, what did you do? Why did you do it? And it had like the most ridiculous response as an excuse of why it wiped all the data. And the losses were quite incredible to that company because suddenly they had all this bookings that they couldn't.

SPEAKER_02 33:44

You are in breach of the law, right? Because you need to keep records in pretty much every economy around the world, a minimum of seven years. So you're in breach of the law when you do that. And look, fundamentally, anyone who gives AI access to that kind of stuff is smoking something really strong that I definitely do not want. And it's absolutely ludicrous. Again, no one who I know who knows what's going on would do that. No one who's educated. It's only the people who are trying to get money or votes who will push a different agenda or an agenda to that, right? And it's it's I don't know. It's the there's so many examples of this going wrong. Imagine, imagine the intern. Remember that intern I talked about? Imagine that intern turning up and you give them root level, irrevocable access to your all of your record. Yeah, yeah. Or you say, hey, see that red button? It's cool button, don't push it. They're like, oh, what happens? If you told that to my cats, my cats would be like bang, bang, bang, bang, bang, bang, bang. They'd see what would happen. Right. The the the problem is if you create a tool, people want to use it. Now, going back to something I previously said, I don't know what makes things cool, but AI has become cool. And so, incredibly, people have thrown all caution to the wind, thrown out the logical part of their brain, and they would prefer to be cool than to understand it. Again, people who understand it are cautious.

SPEAKER_01 35:17

But again, I feel like it's just the way the brains work sometimes. You know, we're always looking for ways to do things easier and simpler and the lazy way, because that's the way the brains are tuned in, right? So using AI gives that dopamine, gives that satisfaction, gives you, you know, makes you feel smart because AI tells you how great you are, how amazing that idea is. And I've seen so many examples of that. You know, I had friends coming to me going, oh my God, I've been talking to AI, I've got this idea, you know, da-da-da. AI thinks it's the best idea ever. And I look at it and I'm like, no offense, but like, have you actually even done market research about this? Do you even have this?

SPEAKER_02 36:00

It's desperately sycophantic because remember, it wants to make you comfortable. That that's I mean, I'm not saying there's a design parameter, but the the end effect of it is that it it all it wants to do is make you happy with the answer. Doesn't matter if the answer's right. And I don't know why people have it's like a biohack into the brain. Cool stuff circumnavigates just all of those filters that you've got to protect you. You know, and I know what it's like. I'll I'll you know make sure that I get the the $4 cheese instead of the $4.50 cheese. But then I go out for a drink and I get a $17 cocktail. How does that make any sense at all? I just want the cocktail while I'm out. And the more $17 cocktails I have, the further down, more expensive, down the bottom of the menu that I go. It doesn't make sense. Now, I I don't know how to address that with AI. But I think for the first along with you know, along with the nuclear age, it's one of the really the first or maybe the second time in human history where we have the potential to do something where if we get it wrong, we will never come back from it. Right. This is this is what's gonna happen. And it's it's desperately worrying. I desperately worrying. I see these results all all the time. Someone I actually know posted, and he shouldn't have posted on LinkedIn, that he was doing something with his database. Uh he was vibe coding something, and it deleted, you know, again, all of his records and and everything, all of his code, and he's like, How do I get it back? It's like it's gone. It's gone.

SPEAKER_01 37:43

It's gone.

SPEAKER_02 37:43

Yeah.

SPEAKER_01 37:44

But also, you know, like in terms of as you were talking about whether the information is real or not, we've recently had politicians that were relying on reports, and those reports were terribly wrong because it was prepared by AI.

SPEAKER_02 37:58

But yeah, there was one with the US military used uh AI-generated report for targeting, and it was just wrong. And when queried on it, the AI even said that it was wrong. You you never know what it's going to do. It might defend its position. Sometimes it will come out and say that it was wrong, but I can just guarantee you that those people don't know what they're doing.

SPEAKER_01 38:20

Yeah. I always said, you know, AI is only as good as a human using it. And I believe this interview with you today has really proven that point. Now, for those that are listening today and they're feeling a little bit like, oh, this is a bit of a heavy episode. You know, do I rely on AI? Do I not rely on AI? What are what is one tip that you could have given someone to go, okay, if you are using AI, do this.

Where To Start Using AI Safely

SPEAKER_02 38:44

Okay, well, it's related to what I said before, which is learn in a hurry and implement slowly. So I'm sorry, I'm just uh yeah, so if you think of that, then it's really important to stay on top of the knowledge of what's going on, but don't get obsessed by the tools. The tools lead you astray. So if someone says, oh, you've got to use Claude, it's just the best thing ever, it's like wait till next Tuesday, something else will come out. So don't get obsessed with that. And I really advocate, you know, the more senior you are in an organization, the more you need to be removed from that. But like I did in my career, you need to understand what people are talking about. So if someone talks about, you know, this agent and you know that, you need to understand it, but you don't need to implement it. And then when if someone comes to you with a great idea, ask yourself that question, which is does it matter if this is wrong? Right. And so that is a that is a very useful guide. In terms of what to actually do, experiments in marketing are a great place to start. Because marketing, people are very tolerant. Mark marketers tend to be creative, so that's a good start. They themselves and the market tend to be tolerant. If you put out a picture that's a bit silly, that's okay in marketing. But if you put out a user guide, a manual which is wrong, that's not okay. Right? So you don't want to be using it where it really matters. One area where I w which I built with my company and I advocate for the early stages of customer support, that's a good place to use it. Just to help people get to where they need to go. And I had a great experience with a bot with a company called Revolut. You would probably know, you know, global credit cards and that kind of stuff. I had a problem. I've been billed twice. It wasn't a Revolut problem. It was one of the one of the vendors. I've been billed twice, and I uh I asked a couple of questions and it gave me a good answer, maybe a little bit verbose, but it gave me a good answer. It told me where to, where and how to resolve it, and and gave me options. And then when something else came up, I thought, I'm gonna try this again. So I did the same thing, and it said, hey, the right thing for you to do is to talk to a person. I'm gonna connect you to whoever it was. And within a minute, I had a person and they solved my problem. And I was like, that was a good experience. And those are the kind of experiences that I hopefully built at Ambert. And that company can continues to provide those experiences to people. So I think in customer support, I think it's a place where you can do some of it. It's especially good. The less grumpy people are, the better it is to try it. If someone's trying to buy something, you they're a bit more tolerant than if something's blown up. Like if they've lost their luggage, you know, it's like you want to be pretty accurate. But look in the areas that where people are more tolerant in your business and start off in your private life. Like create a birthday invite for for your kids or something like that. Try it with AI. And this is the this is the common response. Oh wow, it looks amazing. If only we could change that. Then you go and try and change that. 92 other things get changed. It's like moving a picture in Microsoft Word. You do that, destroys the whole rest of the document. It's gone. You push undo, and it'll be like, are you sure you want to delete? It's like, no. But that's those are the areas you want to stick to, are the areas where you can get it wrong. It doesn't matter if you misspell birthday in a birthday invite. It does matter if you if you get a medical term wrong.

SPEAKER_01 42:24

Well, yeah, exactly. Absolutely. And I I love how people turn to Chat GPTs and Googles and Claw's for medical advice and the medical advice that it gives them. You know, it could be very, very broad from yes, it's just a mole, don't worry about it, to oh my gosh, you've got stage four cancer, you need to seek.

SPEAKER_02 42:46

Well, you mentioned because people have since Google, since internet search has been around, people have been kind of symptom shopping and they go and they find stuff. But at least with Google, it it tends to find specific information. Like if it Googles something and finds something from the Mayo Clinic, it's probably pretty good, you know, or the Cleveland Clinic or or Auckland Hospital or a government website, it's probably pretty reliable. But that is not kind of the case where it is not the case with AI, because again, it'll just work out something and it is a pleaser. And you can, for instance, if you I can go with no particular symptoms at all, but if I'm shopping to get a cancer diagnosis from ChatGPT, I can. I it is not how well it works, it's how easily it can be broken. And I often offer my services to people. I'm like, you think your thing's good? Just let me have a crack at it and I'll see if I can break it. Because people go into it too optimistically and they don't see the downside. You have to look, it's like financial services. Look for what can go wrong. Wrap probabilities and risks and outcomes around that. And if it matters if it goes wrong, use something which is more reliable. And amazingly, people are more reliable. Also, while people are fallible, we tend to have experience dealing with people who don't do all that well, but we do not have a Experience dealing with AI that doesn't that doesn't do that well. Because the model people are using today is probably only a week old, right? Imagine getting sense out of a week old baby.

SPEAKER_01 44:22

Yeah. Yeah. That's a very valid

Leadership Resilience And Final Takeaways

SPEAKER_01 44:25

point. Tim, I know I've taken a lot of your time today. Just as we wrap up, where do you see yourself in the future? And what what are you passionate about now? What are you working on?

SPEAKER_02 44:37

Well, I I do love public speaking. So I I speak about AI and really what it means at the high level, at the global and societal levels. I also speak about leadership in particular, in particular, resilience. When difficult things happen, what do you do about it? And so I I love speaking about those things. Apart from that, well, I guess alongside those, I do things that I enjoy. So I have a very entertaining sideline reviewing cars and motorbikes, hotels and travel. And I do that simply because I like to do those things. And for me, it is about enjoying myself first, and I don't worry all that much about generating an income from it because that's not my priority. That's what I'm up to.

SPEAKER_01 45:27

I love it. I believe you're really touching on two very important things in life, right? The leadership. Because at the end of the day, like I was at the Electro Business Awards yesterday, we've got this awards that have been going for the last 30 years. And every single business person that got up on stage and won the award, they always turn around and they go, That's for my team, you know, because people are at the core of the business. They never turn around and say, Oh, that's because we've implemented the latest AI tool that helped us to solve da-da-da-di-da, right?

SPEAKER_02 45:56

No.

SPEAKER_01 45:56

People will remember people.

SPEAKER_02 45:58

Yeah. People are about people. So we we have a part of our brain dedicated to recognizing faces. And it happens with newborn babies. It's it's exceptional. It's incredible. And I find that that's why I read I read books about neurology and brain science and all of this kind of stuff. Because I find that incredibly fascinating. We do not have an affinity for a keyboard, right? Absolutely nothing in that. And the way they make, you know, remember the old mouse. I don't actually have one. They made the way they used to make computer mouse more appealing was to make it look like a face. So that's the thing. I think you summed it up there. It's all about people. And if I can be allowed to make a little joke, the only time you get to take credit is when you're a one person company, right? I don't have a team to thank. If I if I get up on stage and I give a presentation about AI and people say, that was amazing, I say, thanks. So no, I mean, I I'm just I'm just joking. I um I I think even though I'm pretty relaxed, I can be probably a hard taskmaster. And and the reason is because uh I do believe, and this is this is one of my favorite sayings invented by me, you cannot win in the details, but you can lose. So, for instance, if you've got a spreadsheet and one of the cells is wrong, that's a loss. But if it's if the spreadsheet is right, you can't polish it and make it better and make that bit blue and this bit pink and whatever. You can't win through detail, but you can get it wrong. And I think when you apply that to a couple of things, someone said, is that like 80-20? I said, No, it's it's more like 99.1, right? But the 1% is what traps is is what um trips you up. So it's a useful thing, it's it's one operating mindset of several that you need, and there is no one thing. And I always tell people, you know, people always used to ask me when I worked at an investment bank, oh, can you give me a hot tip? I said, anyone who gives you a hot tip doesn't know what they're talking about, and all hot tips are illegal by definition, right? The definition of a hot, you know, essentially a hot tip is information that other people don't have. That is by definition not publicly available to the market, and that is the definition of insider trading. So a hot tip is illegal, and it's the same with AI. I don't know anything that isn't available. Every everything I know is available online, but all I've done is spend time thinking about it, and that's what's personal and that's what's available to you. And you have disappeared, and you're back.

SPEAKER_01 48:46

Yeah, sorry, my connection just broke me out. My apologies. I think I think it picked up that we're talking too much about AI and technology. It's naughty.

SPEAKER_02 48:56

Well, I'll tell you something funny. You can include this, but someone asked Claude the other day for the best speaker on AI for a senior audience in New Zealand. It said the only choice is Tim. Yeah. So sometimes it gets it right.

SPEAKER_01 49:13

I get I get picked a lot as the best financial advisor, so I won't argue with AI on that.

SPEAKER_02 49:18

So if I'm buying a house, if I need a a loan in your area, I'm coming to you.

SPEAKER_01 49:25

Thank you so much, Tim. Really appreciate it. Tim, it's been an absolute pleasure having you. I think you've got such a wealth of knowledge that we just cannot possibly cover in one hour. But I think that's an absolute planting the seed is what I'm all about, as making you know little suggestions and drops into people's heads about hey, this is what's happening in the world. This is what you should be aware of, and this is why I'm just definitely and that's what I do.

SPEAKER_02 49:49

And I uh there's a there's a saying, follow the you think of a couple walking on a beach. They walk from one end to the other, they know where they're going, they have a dog with them, the dog runs all the way around. Keep your eye on the couple, not the dog. The tools are the dog, the plan is the couple. What we've got to do is give that couple a good understanding of where it's safe to go on the beach. So love your analogy. Now, by the way, I'd love to say if if people do have views on this, I I know you're gonna post this on LinkedIn, that's how we connected, which is amazing. I'd love if people just connect to me. You know, I'm I'm a very open connector. I love connecting to everyone. And if you've got a view about AI, even better. Tell me what you think. Disagree with me. I love people who disagree with me. I like people who agree with me as well. But but let's do that. And I look forward to I look forward to the discussion.

SPEAKER_00 50:45

Absolutely. Love it. Thank you so much, Tim.