Jan. 12, 2023

228 AI Bootcamp for Leaders: Unlocking the Power of Artificial Intelligence for Your Organization with Tom Taulli | Partnering Leadership Global Thought Leader

228 AI Bootcamp for Leaders: Unlocking the Power of Artificial Intelligence for Your Organization with Tom Taulli | Partnering Leadership Global Thought Leader

In this Partnering Leadership conversation, Mahan Tavakoli speaks with Tom Taulli, author of multiple books, including Artificial Intelligence Basics: A Non-Technical Introduction, which serves as the focus of this conversation.  Tom Taulli has been developing software since the 1980s. In college, he started his first company, which focused on developing e-learning systems. He also created other companies, including Hypermart.net, which was sold to InfoSpace in 1996. Along the way, Tom has written columns for online publications such as BusinessWeek.com, TechWeb.com, and Bloomberg.com. He also writes posts on Artificial Intelligence for Forbes.com and is the advisor to various companies in the space. In this conversation, Tom Taulli shares some of the key factors in AI and why understanding the role of AI in organizations will be critical for leaders.  Tom Taulli shares examples of uses of AI, including the role that ChatGPT can play for organizations and in transforming other aspects of life, including education.  Tom Taulli talks about some of the current use cases for AI in organizations and how leaders can start testing to familiarize themselves with the potential of AI for their teams and organizations.  Finally, Tom Taulli shares how nontechnical leaders can stay on top of emerging trends and ensure they are leading their teams and organizations with an understanding of their application in various parts of their business.  



Some highlights

- Tom Taulli on how he became emersed in technology and why he is now focused even more on artificial intelligence.

- The impact of AI on our approach to education

- Tom Taulli on the Venn diagram of AI, Machine Learning, and Deep Learning

- The different applications of AI 

- The potential impact of AI on human emotions

- The opportunities and dangers of big data 

- Data’s contribution to AI bias 

- The benefits and risks of AI in business

- How organizations can use Chat GPT 

- The evolution of conversational AI and where we are headed

- The Impact of OpenAI's Chat GPT-3 on the Future of AI

- How AI will impact cyber security and what leaders need to consider

- The potential impact of AI on industries as varied as healthcare, construction, and agriculture

- How AI could impact the future of employment and the future of work



Connect with Tom Taulli

Tom Taulli Website 

Tom Taulli on Twitter 

Tom Taulli on LinkedIn 

Artificial Intelligence Basics: A Non-Technical Introduction on Amazon 


Connect with Mahan Tavakoli:

Mahan Tavakoli Website

Mahan Tavakoli on LinkedIn

Partnering Leadership Website


Transcript

Mahan Tavakoli: Welcome to Partnering Leadership. I am really excited this week to be welcoming Tom Taulli. Tom is the author of Artificial Intelligence Basics and Non-Technical Introduction. I really enjoyed Tom's book and this conversation because Tom clearly explains the concepts around artificial intelligence, most especially how we as business and organizational leaders can understand their use in our organizations and also the potential of artificial intelligence in the community and society at large. 

I'm sure you will enjoy the conversation. Will want to definitely read Tom's book as well, because as I mentioned in my solo episode at the beginning of January, I believe we are at an inflection point with exponential growth and changes happening all around us as a result of artificial intelligence.

So, it is important for us as business leaders to understand the application of AI in our organizations and the changes that will happen in the community and society as a result. I'm sure you will enjoy the conversation. I also love hearing from you. Keep your comments coming. mahan@mahantavakoli.com. There's a microphone icon on partneringleadership.com. You can leave voice messages for me there. 

Now here is my conversation with Tom Taulli. 

Tom Taulli, welcome to Partnering Leadership. I'm thrilled to have you in this conversation with me.

Tom Taulli: Great to be here. Thanks very much.

Mahan Tavakoli: I love a lot of your content, Tom, and absolutely loved Artificial Intelligence Basics and Non-Technical Introduction because for someone like me who has been somewhat familiar with some aspects of AI in its application, I think your book really helped me understand the various concepts better and try to visualize where artificial intelligence will go.

So, I can't wait to get some of your thoughts and insights on that. But before we do, we'd love to know whereabouts you grew up and how your upbringing impacted the kind of person you've become, Tom.

Tom Taulli: Thanks again for reading my book. That was the purpose, is for the rest of us who aren't data scientists to understand what's happening out there with this new technology. The thing is that in terms of my background, I grew up in Southern California around the Pasadena area, just a typical kid.

He was probably in the early 1980s and my dad, he didn't tell me he was gonna buy me a computer, but he bought one, he put it on my desk and he left the room. And I looked at that computer for a couple days and not sure what to do with it. That was my dad's way of saying you figure it out.

That's the whole point of this whole process. I started tinkering with it and I learned basic. I started to actually create my own computer programs and started publishing them. It was a lot of fun and I just fell in love with it. 

I actually was a big fan of science fiction, so I love these stories about artificial intelligence. But I looked at my little computer there and I knew it wasn't going to do anything with artificial intelligence. I did other things that were a little more practical and that continued through my career, through college.

I started different companies through the nineties. Although I always was fascinated by artificial intelligence I knew that the technology just wasn't there. And really wasn't until the last five or six years that I realized, wow, that this is getting really good. But I thought a lot of the books out there were very technical or very cosmic and I was thinking like the practical side of things.

So, I thought maybe I should come up with a book called AI Basics to help people just understand it. This focus is primarily for business people. But it was funny. My mom's hairdresser, she has a teenager, a 15 year old teenager who happens to be really excited about artificial intelligence.

And for her birthday, she bought her the book and she really liked it. It tells you, these young kids don't underestimate them. They probably know more than you do.

Mahan Tavakoli: Yeah, absolutely. And I'm glad your dad didn't underestimate you by putting that first computer in your room. So I wonder, as the parents are listening, and I'm a parent of a 13 and a 16 year old, what in your view is today's version of that computer that your dad left in your room and walked out?

Tom Taulli: The one thing is kids now are they're digital natives. They've grown up with the iPhone they go to school, last few years. They went to school online. So it is just natural to them, they're not afraid of it. They love it. The TikTok thing they can't get enough of it.

I think that's great cuz it democratizes it. When I was a kid, most kids my age didn't have a dad that would do that. Most of 'em didn't have dads. Most of those dads didn't even know what a computer was or, certainly wouldn't spent, the money it took to buy one.

Now the technology is so ubiquitous that anyone really has an opportunity. And I think that's the great thing today is that it is so widely available, this technology, and then it brings out that creativity. And I know we're gonna talk about chatGPT but one of my editors said that she was playing with it with her 13 year old kid, and she was, the kid was amazed by it.

My editor was amazed by it and I think these kinds of AI systems spark a lot of just creativity and interest and passion. I don't know where that goes, where it leads, where the next Bill Gates or Mark Zuckerberg or whoever, but they're out there.

And the good news is you don't have to be from a wealthy family to do that nowadays.

Mahan Tavakoli: That's wonderful. And you mentioned ChatGPT and I was playing with it with my daughters also. One of the things that they realized pretty quickly is they could now do their essays using chat GPT. But part of what we realized together is that maybe finally it will have an impact on flipping the classroom and the way we have approached education for so long.

We are approaching education the way we did a hundred years ago in industrialization and it's not meeting our needs. So, maybe tools like chatGPT and this conversational AI will force us to reassess how do we develop the minds and bring out the creativity rather than just pure memorization and writing sentences in order to write page after page of meaningless words.

Tom Taulli: Exactly. The calculator was supposed to spell the doom of mathematics. It just really helped with math. The challenge with the essay though usually is, go read a book, go home and at home, write it, and then bring it in the classroom.

And you're talking about rethinking the organization. The other approach is why don't they read the book? They play around on the internet or GPT or whatever they want, and then in the classroom, they write their essay there and then just see how they're able to take that information and come up with, a good paper doesn't have to be a long one, but just, to show that they understand the topic and they can put a sentence together that'll tell you a lot.

So, I don't think it's will spell the doom of the educational system. I think it's a great tool. But if you work it within, this traditional model of the classroom it could be problematic.

Mahan Tavakoli: So chatGPT is one form. A conversational AI but before going much further, I would love to understand, I hear things about artificial intelligence, which has been around for many years. 

Then I hear about deep learning, machine learning. What is AI? What is deep learning? What's machine learning, and how can we think about those in the environment around us and applications in businesses?

Tom Taulli: Yeah. The terms get confusing the way I look at it's just a visual. So, artificial intelligence is a catchall for anything that uses machines to do something intelligent. And, at a very high level, there's strong AI and then there's weak AI.

We're in the era of weak AI. It's using this technology for a certain purpose. It could be for, I want to improve my marketing. I want to forecast to help with my supply chain. That's all weak AI, even though it's called weak, it's still very powerful. But it's for a specific purpose.

The strong AI is the science fiction AI. We're not there. There's this thing called the Deter test where you talk to the AI and you don't even realize it's AI. You think it's a person. Although in some circumstances we probably have talked to AI and didn't know we were talking to AI.

So, I think we've already in that area, but we can't have these freeform, extensive conversations like I can have with another person and I could pretty quickly figure out if I'm talking to an AI or not at that level. So, strong AI is really down the road. That's science fiction.

So we're in the weak AI. So, with this big circle of AI, there's these subsets. One of 'em is generative AI, which is where the chatGPT is, where it creates content. Deep learning is something that is a really sophisticated system to take data, it just takes in a huge amount of data, finds patterns in that data and can predict certain things. 

It could show maybe that an oil rig, a little piece on there, and the oil rig is going to fail. So, it'll alert me before that happens and I can fix it. It'll only cost me maybe 50 bucks to fix but had it not been fixed, it could have cost me a hundred thousand dollars because it would've shut down the oil well. Something like that. 

So, deep learning is just a subset of artificial intelligence. Machine learning is another subset of artificial intelligence. Machine learning is an older, more traditional version more about statistics.

So, if you  remember in high school or college, you had a regression analysis. So I was just looking at, taking data and finding correlations in that data. It could be as simple as saying, if I have three rooms, a bathroom and a pool, and it's in this neighborhood. My house is gonna be worth X amount within a certain range.

That's more like machine learning, a little less sophisticated than deep learning. And then there's make natural language processing, NLP, which is another subset. And that's where we get to chatGPT where it processes our voice, our words, what we type into a computer as a normal person, which is write something or say something and interpret that and get certain results and then maybe, respond to that information.

I think where it gets confusing is there's so many different parts of AI. It's all under one umbrella of AI so you just look at that big circle and then there's these smaller circles, these subsets, like deep learning, machine learning, natural processing, generative AI and so forth that make that whole.

So, that's how I look at it. But you don't have to get too deep on it because really at the end of the day, it's taking data, it's applying some math to it, an algorithm, and getting the results. Basically that's what it's about. Now, it can get really complicated and you can have all kinds of really sophisticated analysis, but that's really the concept.

Mahan Tavakoli: So, the functionality of what you just mentioned, taking that data, applying some math to it and getting some results, we see it all around us in the world. In your book, you start out with the example of the different forms of AI that Uber uses just in the simple app that we use on an ongoing basis whenever we use Uber.

So, what are those applications? Just so we start understanding that AI is not this vague concept. It's already in our lives all around us.

Tom Taulli: Yeah. Uber is a great example. Most people have used Uber. It's actually a very complicated math problem to figure out, the demand for something and availability of drivers and coordinating all that. It really, at the other end of Uber is a really sophisticated AI or machine learning or deep learning model that takes all that information and makes everything work.

TikTok, that's a little controversial right now because of China's involvement, but it's uncanny because it seems to always give you videos you want to watch, but what it does is it basically the data that is created from the viewing habits of its users, from you and other users, or users like you and then it will then analyze a certain video and see, does it fall within certain metrics or certain criteria? 

And they'll say, yes, it does really well. Let's show that next video to that user. And before you know it, that person's addicted. So, TikTok is one of the most sophisticated AI systems at scale, and it's real time. That's the other thing too, it's one thing to just take a lot of data, do some analysis, and come up with some results. Try doing that in real time. Again, with Uber, try doing that when there's someone outside waiting for a car or someone that's just sitting around looking at their phone and looking at videos, that's tough.

So, we're talking about cutting edge AI, but you as the user shouldn't have to know why that's AI or not. It's a great experience and there's some magic behind it, that's the AI. But you really don't need to know it. But it's all around us, definitely.

Mahan Tavakoli: Now a couple of concerns around that. Trying to understand it, Tom. Yuval Noah Harari talks about the concerning element of the AI, knowing our emotions and being able to manipulate our emotions better than we can. And there are AI companies working on assessing emotions of users and are doing a great job with that.

So, what are your thoughts with respect to the interaction of this AI, whether it's on TikTok, Facebook algorithm, or other uses of it where it's very effective at determining human emotions and then guiding those humans to do what the programmers of the AI or the people using the AI want the humans to do.

Tom Taulli: That's a really good question. Something called a mode of AI or that is really focused on that. I talked to someone at Facebook, he told me that, when you have 2 billion users to work with, it is a great playground for your AI system, cuz you're a big piece of the global population across so many different countries and cultures and so forth.

But he said, despite all the cultures and differences and whatever, he says it's so remarkable how predictable human beings are. You collect all this information and you see these common patterns through the data about how people are, how they react, what they do. People tend to just basically, they tend to be the same person they always have been.

People generally don't change. They like certain people. How is it that something like Match can figure out who you like or you want to date or eventually marry based on just, you would think not a lot of usage of that app. But it's because we have DNA, we're wired certain ways and we give off these patterns.

In terms of the dangers of this data scientists tend to be white, tend to be from wealthier families, obviously highly educated, so they bring a certain vantage point to their work. I don't think they go and look at us, model and say, we're gonna do this devious thing, and, something diabolical, I think what happens is there's a lot of unintended consequences to what these models do.

Now some of it is very intended, these companies want to create engagement. Their goal is to keep people coming back and to keep using the app and that does raise some troubling issues. But a lot of times the issues come about just from unattended acts.

So, there's a famous case a few years ago when Apple launched its credit card or debit card, with Apple Pay and so forth. And quickly what happened was Twitter lit up because it was providing higher credit limits to males versus females. And even Steve Wozniak, the co-founder of Apple, said he had a higher credit rating even though his wife had a higher income than he did.

And he didn't really work. He just, he's being Wozniak and speaks and does that kind of stuff. The thing is apple has some of the best data scientists in the world. They work with Goldman Sachs, which has some of the best data scientists in the world. And they said, you know what we never included gender in our data as a determinant.

We excluded it from the data set. But what happened was gender can be revealed by other types of data. Maybe certain professions that are, tend to be more female oriented or male oriented, and so that was the problem. So they didn't go in saying, we wanna make sure men get higher credit limits.

But that's what happened because the models had this bias in the data and gave off certain results. So these are tough problems. Data scientists are generally pretty ethical people, just I think most people are.

But just like everyone else we have these latent biases that we don't realize we have.

Mahan Tavakoli: It's the unintended consequences, as you mentioned, and the scaling of some of those biases that might be inherent in the data. Brian Christian also uses the example of AI that was used in Amazon for recruiting, and pretty quickly they found out that it was biasing for certain experiences people had, which would primarily end up advantaging white men that played lacrosse, for example, or went to certain schools. 

So, part of what that says though is that the data sets that the AI is working off of is really important and these are cases that have popped up and we see. However, all around us from applicant tracking systems that a lot of organizations use to screen resumes to other parts of the corporate world AI is being used. 

How can organizations think about it in a way where the data set is a data set that doesn't scale bias?

Tom Taulli: Part of solving the problem is recognizing that there is a problem. Years ago, we didn't have these discussions. So, I think it's good news that we were having these discussions bring it up. Chat GPT even in its results, it'll even say some something like, yeah, I'm not even too sure this is the correct answer, or, be careful. I think just recognizing that there's potential problems is a good first step.

Now that's a small step. The other is that when it comes to AI, we always talk about the cool deep learning, cuz they sound fancy. These algorithms, most of them are actually open source. They've been around for a long time and anyone can go pick those up and run 'em.

The hard part is data. And sometimes it's even mundane things like there's repetition in the data or there's data missing, or data is out of bounds or whatever. So, there's this thing called data wrangling where you just go in there and just fix the data. So that's the first step.

The second is, and this is where human capabilities are so important, is to look at that data and say, is it giving off certain indications that are skewing it certain ways and you can start testing for that. So I think, just the fact that, it's become a problem, companies have hit their reputation has been hit to some extent by this is potential legal liability, they are putting in some of these guardrails in there and trying to figure out ways to minimize it. But, there's no perfect solution here. But at least I think companies are being more responsible about it and that's a good step.

Mahan Tavakoli: Which is also why Tom, it's. Even more important for all of us, especially business decision makers who are not necessarily the technologists to understand AI, the issues and its application, to be able to ask right questions in order to determine the use of AI.

So, I think it makes it even more important that it is not just something that a few people in the organization understand and are running. Part of what you say in your book is identifying which business problems to solve using AI and assessing what value will be created are critical for the success of all AI projects.

So, as CEOs and executives in organizations of various sizes, as they're thinking about their business issues and business problems, what are the ones that, at this point, lend themselves most effectively to testing application and understanding through artificial intelligence?

Tom Taulli: I think the first step is to just evaluate your company and look at your processes and so forth and don't do a custom project. Look for off the shelf products first. You mentioned chatbots. There's a lot of chatbots out there that do quite well, that you don't have to go and create your own chat bot.

In fact there's hundreds, maybe thousands of different chatbots on the market. The hard part is just trying to find the one. But so you could get that chatbot, put it on your website, and then, start to train it for certain things that are particular to your business and then evolve that. 

So, I think the idea is to look at a specific part of your business and then see if there's off the shelf technology that you can use and maybe you can adjust and tweak it and improve it over time. And that'll be your, kinda your first step down the road to AI.

Now over time, as you build those muscles, maybe you can do your own custom project. The other thing too is, looking at where do you have your data? Where are you creating a lot of data in certain areas? That's another place to look for AI. If you have salespeople and they're creating a lot of data, how can you use that to improve their performance and their productivity?

So, there's AI products like Gong for example, that will take that information about like Zoom calls and communications with customers on CRMs. Take that data and figure out what works and what doesn't. Certain messages might work with your customers and certain ones don't.

That can be extremely valuable information to know. Salespeople are just trying to do what they can do, but they don't necessarily know what's working or what's not, but the data can. So anywhere you have a lot of data that's another rich area to explore for AI.

And if there's areas that really require a tremendous amount of judgment or very serious and very important, try to hold off on those areas. Go for parts that the low-hanging fruit, the easy areas, and then you can work your way up to the more sophisticated areas.

The other thing too is you may wanna start more in the back office than customer facing because if you mess up with your customers on something, you don't wanna lose a customer so that's another thing to think about.

Mahan Tavakoli: I appreciate those recommendations, which is look back-office first. You don't necessarily need to develop your own. There are a lot of great off the shelf products, and you also mentioned in your book data is the lifeblood of AI. 

So, you look at a place in the organization where there is a lot of data, which then provides the opportunity for AI to find those patterns, which then you as a human executive can gauge whether that's of value or not and how to adjust it. So, I really appreciate those thoughts on the application of it. 

Now, for me, Tom, I was shocked at the capability of GPT at this point, knowing that they have the next version ready, which is supposedly five times even stronger than this one and thinking about its applications in organizations and the potential disruption in a lot of different jobs and roles that we currently have. 

Would love to get some of your thoughts and perspectives. But I know it has opportunities for improvement, but at this point, this level of conversational AI that chatGPT has, how can it be used by organizations and what are the types of disruptions in our organizations and business world as a result of this capability?

Tom Taulli: If you're new to AI, you don't want to do custom projects. The first step would not be to use chatGPT for your own business, per se. But it's probably next year it'll get even exponentially more powerful than it is today, which is scary thought to think about. I think the GPT three has 175 billion parameters in it. The GPT 3.5, we're not sure how many parameters are in there, but it's obviously more.

ChatGPT uses GPT 3.5. Version four, which could come out next year, could be a hundred trillion parameters,

Mahan Tavakoli: Oh.

Tom Taulli: It could be that high. That's mind boggling. If you think it's good today, it's gonna be a whole lot better tomorrow. And that's the thing about this technology.

It just only gets exponentially better. So now I saw one use case where a doctor always has to write a letter to a health plan to justify reimbursement for a drug or some type of treatment.

You need to write in a certain format, you've got to include some reasons and also citations of why this should be used and why it's a proper procedure. This doctor gave a two sentence description of what he needed and chatGPT came up with what looked like a flawless letter, for the health plan with citations for this. 

And it was a procedure I never even heard about. It was one of those crazy things. Unless you're in the medical field, you won't even know what it is. So I've talked to some financial planners. He said, I had a client whose dog died and I wanted to write a letter to that person, or an email of consolation. So he put it in chatGPT, and it was actually the perfect consolation letter that you could have for a client. 

Those are things, the touchy feely type of things, which are very important in the business world, this could actually be a really powerful tool, where you don't need the facts because the accuracy is questionable in some of these things where, and you're not even too sure when it's accurate and when it's not.

Cuz it won't necessarily tell you. But some of these, soft skill things chatGPT could actually be really effective for your business. 

Mahan Tavakoli: I agree with you, Tom, and I laugh about it because as you say, chatGPT doesn't give you a certainty level on how sure it is of an answer. So on some things it can be wrong and it can still give you an answer. 

But as you mentioned, it is great on the soft skills. So here we are that AI in instances is better at soft skills than we are as humans.

Tom Taulli: Think about it. If I were to try to write an email to somebody who lost their dog, I wouldn't even know what to say. I, where do I begin? And it happens at my computer, just really good at it. And the other thing where it's been used is on like Tinder. So, there's a match between two people, but you always wanna have that first, something that's really engaging with the other person, but that's actually tough. 

But people are doing are taking information from the profile, putting the chatGPT, and coming up with great openers. The thing is that that person you think is who that person is really just an AI at the end of the day, in terms of what they just said. It does raise all kinds of interesting issues and societal things, it is just opened so many things.

It's hard to fathom. But yeah, I think the soft skills part has been underestimated. But again, that can be some of those important parts of business. So I do think there's actually use cases there, but it may not be the use cases.

You would think at this point down the line when the accuracy gets better or at least you can determine the accuracy. We can start using this for more mission critical applications, but I don't think it's there yet.

Mahan Tavakoli: I've also seen some good uses of it in content creation. Not superb content creation, but a lot of content that is put out typically is not superb anyway. And I've seen people that have redone websites and gotten the websites to become a lot more effective in the sales message or whatever they want to communicate, so there are those uses for it. 

So if we fast forward this capability and this is just open AI. Google has a deep mind that they haven't opened up yet . So if we fast forward over the next few years, this level of conversational AI, which in my view gives more people on an ongoing basis, the ability to interact with the artificial intelligence tool rather than just organizations using it. 

How do you see the evolution of this conversational AI over the coming few years?

Tom Taulli: I do think it's probably on par with what we saw with the Netscape, the internet revolution. PC revolution during the eighties. The iPhone revolution. And the thing is when those came out, probably a lot of people didn't think, what the use cases and applications would ultimately be.

And I think you brought a good point about, computers being much better than us at writing content that's very creative and engaging. The thought of always has been that AI is going to automate lawyers, who write boring contracts and now it looks like if you're a commercial artist, you should probably rethink your profession.

If you're writing poetry, maybe think about something else. Although poetry never really made a lot of money least for most people. But it's these creative fields that are under the gun because this technology is kinda the point where it's just very good.

And it's just the start. We're, we just started on this journey. It's just gonna get better and better. Where we're gonna be in a few years, I would say that the people at OpenAI don't even know the answer to that question.

But they've been clever because I think they realized this technology got to a certain point where it was very general purpose, good and effective then they launched it. Cuz it could have been a huge fail. I think that the timing was just about right for it.

And now it's just up to our own creativity, which is amazing to take in directions. We just never thought it would go. These things that we're talking about today when I was writing this book, I never thought of that. Things like that would happen. 

Mahan Tavakoli: So it has a lot of potential that we can't think of at this point, but I had a conversation with Azeem Azhar on the exponential age, and in my view also, Tom, we are hitting the exponential curve with it. So, you mentioned and I agree with you, whether it was the introduction of iPhone or Netscape it is similar to those, but I imagine because of the exponential curve of this, we are going to go through the changes a lot faster. 

And that's one of the things that I think will have a significant impact on a lot of aspects of our society as we transition, which is why we need to think about it and do it right now. One of the concerns that I have, Tom and I've heard from people is the fact that OpenAI has put guardrails on what this artificial intelligence can do.

You can't ask it questions intended for nefarious purposes, and people have tried workarounds to it, but in general it has its limitations. But those are limitations that OpenAI has put on it. There are other AI platforms. There are others working on AI. 

How can we make sure that AI is used for good rather than used for bad because not everyone is going to put the same guardrails. And how do we make sure that the right guardrails are put on there? Sam Altman and his board decided to put these guardrails, someone else might decide to put other guardrails. 

Would love to get your thoughts on that.

Tom Taulli: Yeah. There's the absolutist position, the absolutist who just says everyone should have whatever they want, do whatever they want. The thing is, once technology is out of the bottle, you can't put it back in. There's things like these transformer models, all this stuff, people know how to do 'em now.

So you can create open source systems that do what chatGPT does and we're gonna start seeing those in the next few years and they probably won't have guardrails. How do you stop that? You can regulate it but the regulation just takes time and consensus and cookies have been around for how long? 30 years. 

And it's only been the last five years that they're regulated to some extent and all the regulation is, do you accept or not accept on, the cookie that's the thing. Technology just will continue to get better and will be free and just like we've had to learn how to work with social media or any of these other things, we're just going to have to do the same with AI.

Just because it came from a computer doesn't mean it's correct. Some of the first key use cases I saw were people using chatGPT to create malware and viruses. And it actually did really good jobs in creating those and finding zero exploits and things like that. Then they finally caught on and they actually put a stop to that and put their ways even around that. The bad actors have an advantage because they have time on their hands and there's a limited number of people on the other side.

It's basically the cyber issue. It's an ongoing arm race and businesses are going to have to invest in those types of systems. And it'll just be a cost of probably doing business going forward,

Mahan Tavakoli: Which is why it makes it really important for all business decision makers to understand the applications, the potential vulnerabilities. Now, I would love to get your thoughts. You also share some in your book with respect to application of AI in cybersecurity, construction and talent management.

Tom Taulli: So cybersecurity, I think AI is going to be extremely important. Because there's a couple problems. One is it's hard to find talent. The talent shortage in cybersecurity is off the chart. And the other thing is the threats just continue to increase, and the ransomwares is off the charts, or phishing.

A lot of these are just routine cybersecurity threats. So I think AI can do a really good job of automating a lot of what has been the old ways of doing things. But I think what will happen is at 10 years down the road, it'll be automated protection systems against threat actors going against each other.

And they'll be like, no humans. It'll just, it'll be out of control. So there's someone other AI attacking my AI. So I do think at some point it's gonna be automated against automated AI, against AI when it comes to cybersecurity.

One thing about cybersecurity that is the unending growth industry that there ever was one, and it'll grow till I die and everyone else dies, it's not going away. So if you're looking at a very, secure career path in our technology world, cybersecurity is probably number one.

Construction is an area that obviously is a huge area. We have the infrastructure bill that will put even more money into it. But if you look at the technology spend and R&D and tech and construction is very low compared to other industries. But they have this kind of a similar problem to cybersecurity.

They have labor issues. It's hard to find qualified people, and yet there's a need to build these things. I think what we're going to see is AI come into the market in a bigger way to help with that automation. It could be also to help maybe with safety the design or digital twins.

A lot of opportunities to automate when it comes to construction. And in terms of talent management you mentioned some of the issues there, with Amazon and their HR system they had to shut it down. But they tried multiple times to get it right. But the ultimate problem was it was a data problem because most of the people who were applying for those positions happened to be male.

So it was pretty much a biased sample. So I would say on talent management gender devAI or this chatGPT could be very helpful. Just in terms of really basic things like HR manuals does anyone really update those things or read 'em, I don't know. But you can use generative AI to update them in real time. If I'm an employee, I can ask it a normal question and I get a normal answer. Just things like that could be really helpful. 

Any way to reduce a lot of the friction in the process of hiring, to speed up those cycles can go a long way. And a lot of that has to do with the paperwork and just the stuff that you just have to do. I know employers are getting better at it. But still I think some of these HR systems have a just terrible. 

I do a lot of kind of like contractor work and I'll get into their HR system and I can just look at it and I just say, there's gotta be a better way than the way they're doing this right now.

You still have to be careful about how you source the talent. And I think you can get overly reliant on the technology and miss out on quality people. Their resumes may not have the kinds of things that stand out, but they may still be great employees. And sometimes a great employee is about passion, the willingness to work, excitement and resumes don't necessarily tell you that.

So there's ways of really enhancing talent management with AI and computer and software technologies. 

Mahan Tavakoli: There's a lot of potential and I know there are AI tools that even do the video interviewing at this point. 

Now, I would love to also get your thoughts, Tom. We could do lots of episodes about the future of warfare as a result of AI,

Tom Taulli: Yeah.

Mahan Tavakoli: But at this point it sounds like there's an arms race on AI within the US and China would love to. Are there any other major players and how is the US-China relationship playing out in this AI arms race?

Tom Taulli: Europe is another block that's been trying to invest more in AI. You can even look at Canada. A lot of the innovations in AI have come from Canada in the last 20 years. Jeffrey Hinton and these great universities that they have.

One thing about what Covid has taught us is that we're not the only game in town. There's smart people all around the world and China, they train so many engineers and they're investing a huge amount into ai. So yeah, the two main powers when it comes to AI are the United States and China.

And the relations between the two countries, seems to be, guess it's, maybe it's stabilized to some extent. Clearly it's kind of a bit of a cold war between the two, and that war is more on the digital side. And one thing which is also often missed about this is that AI requires really sophisticated computers, not just software.

You gotta have a computer to run the software. So, Nvidia is really the main player and hardware chips and machines GPUs, graphics processing units. There's already been moves by the US to prevent technology transfer between companies like Nvidia to China. And there's probably gonna be the reverse too, and then there could be the showdown over Taiwan because most of the world's semiconductors are made in Taiwan.

So, I think that the biggest threat really is what happens with Taiwan and their chip making capability that could set the United States way back economically, not just with AI, but economically if there is a problem in Taiwan. So from the software standpoint there's less of a national security threat.

I think that the national security threat is really on the hardware side, chips and these GPUs and these machines that run the AI.

Mahan Tavakoli: Would love to know about the potential impact or what some people call technological unemployment. Andrew Yang ran on this platform. There've been a lot of people from Silicon Valley talking about universal basic income, and they have closer to a doomsday scenario, not exactly doomsday scenario.

In essence, saying over the coming years, we need to transition because AI is going to take all of our jobs. There are also techno optimists that say every time that has happened with cars, with steam engines, people say no one will have jobs, and new jobs are created. 

We'd love to give your thoughts on the impact of artificial intelligence on the future of work.

Tom Taulli: John Maynard Kanes, the Great Economist wrote a paper in the thirties where he called it the 15 hour work week. And he says his grandchildren would have a 15 hour work week, which is interesting cuz he was gay. They didn't have any, he didn't have any grandchildren.

But NPR interviewed his brother's, grandchildren or sisters, and then they were all working 50, 60, 70 hours a week. But I think he was onto something his idea was machines and automation would create less of a need for human labor.

I do think that this chatGPT has shown us that we're very predictable as people. We may not be as unique as we think either and I think part of this whole debate is our self delusion about how great we are as people and these machines, and I've done stories about this and Oh, they'll never take my job, yo, it, it could never, give you copy or whatever, tell you, hold your hand or whatever.

There's a lot of truth to that. And I even, I talked to someone who's in Silicon Valley, who's a top founder of a company, and he told me that the jobs that will be the last to go are those like nursing. He said a doctor would probably be automated before a nurse because what is a doctor? A lot of times a doctor just does some tests and prescribes you some medicine and a lot of it could be automated.

Whereas the nurse actually has to take care of somebody. And robots, famous last words, but robots will take a while to do that. So I do think at some point a lot of what we do as work and for labor is gonna be automated away. I don't know when that time comes but I do think at some point, given the power of this technology, it will happen and businesses are gonna make it happen because if I can use a machine to do something, or I hire an employee for a hundred thousand a year, and it costs me a couple hundred bucks to do this on the machine.

I'm gonna do it on the machine. So what that means for society is we'll probably have a lot of free time in our hands. The other scary part is there probably will be a concentration of wealth and a very small number of people, and the rest will not. And there would probably be an element of this is not a fair society.

But I think that's quite a ways down the road. But I do think it'll happen at some point.

Mahan Tavakoli: So it's important for us to have these conversations and in the interim it's important for us to constantly learn about it and its application in our organizations. 

In addition to your book, Tom, and following your work, are there any resources you typically find yourself recommending to people who are super technical AI

Tom Taulli: First chatGPT. Ask it a question. I've asked different AI questions. Gets it pretty good. The other thing too is, some people are visual learners and YouTube is a great tool. So if I wanted to know what deep learning is or what hidden layers are a great place is just go to YouTube, type it in, and then.

Some of these have really cool cartoons and descriptions. It really brings it to life. So I think YouTube is another good resource. There's online courses like at Udemy, if you wanna go deeper into it you can do things like that.

And the courses might be 20 bucks or 10 bucks, not too expensive. So there's more than enough resources out there to learn about this, that's definitely the case.

Mahan Tavakoli: First and foremost, reading your book and following your content. Tom, where can the audience find out more about you and your work?

Tom Taulli: I do have a website. Just put my name, tomtauli.com. And it has links to my books there and content. And you could go to Twitter @tauli and anytime I write something I'll post it on Twitter, so you could always follow me that way too.

Mahan Tavakoli: I really appreciate it. Tom, I enjoyed your book a lot. I've loved reading the articles that you've posted on your website and the content you create and. The forward that you have by Sridhar Bamboo, co-founder of Zoho at the beginning of your book, captures the essence of it that AI, if handled correctly, can act as a sweeping democratizing force. 

It has and will eliminate from our lives the drudgery of the past and free up a tremendous amount of human energy and capital. But that if is far from certain, AI executed irresponsibly has the power to destabilize large parts of the world economy by causing as many people fear as shrinking workforce, reduced purchasing power for the middle class and an economy without a wide and stable base fueled by an endless debt spiral.

So I really appreciate your writing, Tom Taulli in artificial intelligence basics and non-technical introduction and the articles that you've written that help more of us, including business leaders, understand AI so we can apply it for good in our organizations and community. Thank you so much for joining this conversation, Tom Taulli.

Tom Taulli: Great. Thank you very much.