AI's Boardroom Revolution

Both of you are very heavily involved in NACD , the National Association for Corporate Directors , and you've just written a fascinating book called Governance in the Age of AI , and this is a director's handbook . This is meant to talk to directors of companies about how AI will impact the governance of their organization , but it has so much more legs than that .

It has just an expansive view of how AI is going to affect decision-making and how to control the environment for AI if it's controllable at all to be able to figure out how companies can do well in that environment . So , gentlemen , welcome .

It's even rarer that you have a group of directors that feel like they're knowledgeable or informed on the subject , and that was one of the drivers that Ashwin and I got together and collaborated on this book , because there is a glaring gap in the boardroom for each director to become challenged , to become more knowledgeable and informed about not only the technology

itself , but what's happening . We're not trying to make every director a technology expert . That's virtually impossible and it's not warranted but what we're trying to do in this book is to describe what the possibilities are and also what the risks are as we move from artificial intelligence to general artificial intelligence and beyond .

determinant of corporate decisions . What the hell do we need to pay CEOs ? Maybe we should pay AI instead of paying millions of dollars to CEOs . And since you're a CEO , Ashwin , I'll start with you .

I think it's a misnomer in some ways , because what it is is a complex mathematical engine that is using input data to generate the next quote-unquote token , which is nothing more than a piece of a word , a word , a phrase or a sentence .

To attribute intelligence to that mathematical engine would also presuppose that that engine has the ability to discern and make decisions that are value-based . It has none of those . It's a pure mathematical engine . So you know , it's like giving a calculator 2 plus 2 and saying equals , It'll give you 4 , no matter what you do .

So this is not a dissimilar engine in that regard . Where it's a , it's , I could almost call it a prediction of the next token engine and not really artificial intelligence .

Because I think when we use the term intelligence , we are also connoting that the intelligence has the ability to discern between right and wrong , good and bad , ethical and non-ethical , legal , not legal , formal , informal .

In other words , they take off into la-la land and make up stuff that doesn't exist because they don't know that it doesn't exist .

You can tell it to look at formal literature and ground itself in that formal literature , but it doesn't always obey because it has a mathematical engine that is asking it to predict the next token that is most likely , asking it to predict the next token that is most likely , and it's the likelihood that it now forecasts as the next sentence and doesn't necessarily

go back and say I need to be rooted in reality . Now , if you're planning a party and you ask it to give you five ideas which you haven't thought of , it is wonderful at doing that . We don't call that hallucination , we call that imagination . But when you ask it a formal question and it imagines something that doesn't exist , we say it's hallucination .

So we can't have it both ways , you see . So I think we need to be careful about when we call it artificial intelligence and what we attribute as the characteristics of intelligence , before we agree that this is artificial intelligence .

So if you're a director in an org , in a company , and your CEO and management team comes to you and says hey , you know what , we've looked at the business , we've analyzed it , we've had the AI analyze the treasure trove of data whether it's relevant data or irrelevant data and it tells us that the patterns are stacked up this way and so we're going to change

strategies . We're going to go , because the AI pointed this out , we're going to go , change strategies and then the strategy doesn't work , how do you avoid that from ?

You might find this interesting , joel is that outside theS there are boards now giving AI a board seat ? Oh , my goodness or a board observer position which you know you mean a formal position . Yes , so that has started Now . I'm not suggesting that ATT is going to jump at that point .

Artificial intelligence is a board member today , but you can see where the possibilities are . Maybe they'll have better results . It's possible . It's certainly possible , but the future is unlimited . As Ashwin indicated , these are machines , but you can see where the next generation , after generative AI , may be going .

It's general intelligence , it's super intelligence and that talks in terms of equaling human capabilities or surpassing human capabilities . That's way beyond the scope of what we wrote . What we wrote was a handbook for board members and executives to get more knowledgeable about this generation of AI , which is generative AI .

They're very excited that in one place they can see a lot of this basic information available and we have at the end of each chapter , chapter , a summary of some of the key areas , key takeaways , key questions that board members can actually take that into the boardroom and ask thoughtful , intelligence questions about a topic that most of them are not experts in .

, are saying , yeah , this is going to happen , it's happening now . So how much rejection of the premise that AI is going to be influencing strategy ?

So , if you look at historic data , which most companies have in the plentiful supply Not necessarily representative , but historic data so the historic data tends to have patterns that may or may not be hidden , and the first generation of AI was all about discovering these hidden patterns that were difficult for the human eye to perceive .

So in some ways , it's applied analytics , but using AI to provide those second and third order patterns that were previously invisible . From there you go to predictive AI , which has the ability to look at interval data and predict what the next value in the interval is going to be , whether it's a time interval or a value interval .

I think we went from there to the token generation mechanism , with generative AI and the transformer mechanisms that gave you the next word , the next sentence , and so on . What we now are seeing is the fourth generation of AI , which is agentic AI , which has the ability to take input data good , bad or indifferent and make rudimentary decisions .

So it's particularly powerful when you have workflows which are repeatable , with patterns that are recognized and automatable .

Is it and I think this is a point of confusion on the part of the public in general when it figures out what it is or how it is it's going to do the next thing , or how it is it's going to do the next thing ? What is it ? How broad is the data that those decisions are being based on ?

Is it just what it's seeing within a company or is it broadening the lens out to include more data from other areas ? How are companies dealing with that ?

So you have an underlying large language model that's been trained with anything that's been thrown at it , but it becomes the basis of the container , and then you further augment the container contents with company-specific data and then you say this is the workflow , here are all the systems that are involved in the workflow .

You provide access rights to the LLM and this agentic model and say draw this information from this system , this other information from this other system , so on and so forth . I mean , imagine , I would argue , 50 , 60% of routine work in a corporation is routine work .

There is not much value added beyond searching for information from multiple sources , arriving at a conclusion and saying this is the most likely next action . Those are being automated with agentic AI .

But it doesn't get to the core of your last question , which is strategy , because I think strategy is a value-based outcome-seeking mechanism where you're looking at the corporation's mission , its long and short-term objectives and constraints and trying to come up with the best next likely thing that it should be doing as an organization To win , to win right .

The whole point here is to have an outcome that is better than what you currently have . So to do that , I think you need human intelligence . You could have an AI machine listening , as Dean said , providing you with an input , but it cannot be the determinant .

I mean , is there a credibility issue , both with the media and with employees , if you're saying I'm making this decision based on AI ? Is that an ?

, that's not good news . So the fear factor is real and in fact , when we talk about generative AI , we have management from the HR department on to other executives , coaching us to let's not scare people about their job security . So the threats are real . I think management's very aware of them because they hear them every day .

But what we see in the boardroom is the fact that sometimes senior management the executives will get very cautious because they see the same script going forward . And that is what is their role going forward .

Is the board going to take the information we provide them , come up with their own ideas and they go off in a different direction , and then it questions what's the values of their decision ? So you've raised a very important point . We're in an evolution . This is not turn on the switch . Everybody moves .

The generative AI Companies go at their own pace , depending upon a number of internal factors that are unique to the organization , but sometimes it gets down to human fear about how you know what's how job is my ? How secure is my job ?

How much does it depend upon actually interacting with other humans and maybe other AIs to be able to influence or cajole or , you know , create some kind of enthusiasm ? And the other is how formulaic or non-formulaic the job is . Is there a wiggle room , as you were talking earlier about ? Most work is routine or follows a routine pattern .

Most work is routine or follows a routine pattern . The more routine it is , the more formulaic it is , the less , the more likely it is to be gotten rid of by AI . Have you dealt with , though , the other element of corporate management fear , which is shareholder lawsuits ?

But if you look back in history , there have been many instances of new technologies that have influenced decision making , leading to lawsuits resulting in tort being reformed , reflective of the new normal . I think we'll see the same pattern here . The stakes are high , there's no question about it .

And to Dean's point , I mean , last week I addressed a company of 100 people they're in the data business and their HR head to Dean's point said make sure that you talk about this as an adjunct and not a replacement technology . So you know that guidance is in itself a caution because the inevitable seems to be written boldly on the wall .

People are recognizing it . They don't have an answer to it . They don't know how to navigate the legal rules because there aren't any yet , but they can see that practically it makes a lot of bottom line sense for them to go toward it . But it also is a democratizing function .

So it could well be that there is a deflationary pressure on the top line , because if you can get the same outcomes that you today are able to with a fraction of the number of people , which is a massive part of the cost in an , you know , especially asset-light corporation . Why do you need the people ?

Now that information comes from various sources Some of it's proprietary , some of it's very confidential that gets out into the public work stream , it becomes part of the training data that's used and all of a sudden it surfaces in an unpredictable way and you can imagine , with creative attorneys they would certainly pursue that as damages .

So I think my own view is , if I was going to ask to predict , I would say that the attorneys have a rich field ahead of them in terms of what the possible litigation opportunities might be .

What would that mean have meant in my career ? Well , I have a bachelor's in accounting . I have a master's in accounting . I have a master's in taxation . I was steeped and educated on tax accounting principles and I felt like I was very competent and fortunately I landed with one of the best firms in the world , pricewaterhousecoopers .

If I was to play that over again and if I was 23 , thank God I was 23 . But if I was 23 again , I'm not convinced . If my goal was to become a partner at a firm like PwC , I don't think doing my path is the right path . I think it's more in critical thinking . I think it's more in data analytics . I think it's more in computer sciences .

It could even be in prompt engineering . You know what's wrong with becoming a prompt engineer who becomes a partner at PWC ? I think partners will exist at PWC . Partner at pwc . I think partners will exist at pwc . How they ?

As a professor I have at chapman , I have a really interesting mix of students and because chapman has one of the best film schools in the world in the business college , where am we tend to get a mixture of students that are taking business courses but they're film majors , or vice versa . Recently , I had a student come to me and say how do I get a job ?

I am an AI systems designer . Well , what that really means is he has the imagination of a film person to figure out how you could actually apply AI at a particular situation , and I'm thinking that to your point . About what would you advise young people to do ?

This is something the world is going to need in the next decade or two are imaginative people who are not necessarily classically business trained to be able to just use their imagination to figure out okay , well , here would be an interesting way of being able to apply it .

My worry is that I wonder whether the HR departments of larger companies are open enough to this kind of redefinition of talent . What's your sense on that ?

It could very well be in one of the traditional businesses or traditional industries that are looking for creative people who have this new , that have a view and a skill set that's very transferable to what is going to be needed in the future .

Indeed , I mean A you have the indoctrination and you know there are ideas that you can't challenge , whether they're free market ideas or Marxist ideas , but there tends to be sort of orthodoxy , and I think that's a problem .

Plus , I think a lot of creative ideas come from knowing history , knowing religion , knowing literature , and I'm not sure that many of these kids do it at all . I mean , I am astounded by the books they haven't read . I mean like , for instance , I don't think you can have this discussion in terms of societal impact unless you read Brave .

One of the things that you can do is when I look at an economic situation , I'll go and I'll say , well , this happened in Rome at this time , this was something that happened in Germany in the 1920s , whatever it is , these kids don't have any of that . Actually , the AI is probably more likely .

You want those allocators of resources to be very deep in making those bold decisions which can sway the scales of history , so to speak . Right , I mean ? We are seeing a world that seems to want to skim on TikTok and move on to the next day , and that's not a good place to be , because that's very dystopian . I mean , tiktok is nothing but an algorithm .

It's a deep algorithm . It's been trained by somebody who's clearly in the game to make money from it , and the more it absorbs you , the more clicks you get from TikTok . If you are the owner of TikTok , you're making a bundle of money .

Why am I paying this guy when I can get it directly from AI , as opposed to somebody ? When I think of the really great minds I've known and I've known a lot of them , and you can even start with Daniele Struppa at Chapman At Chapman , he knows Greek , he knows Latin , he's well-read . He knows history , he knows art .

He knows Greek , he knows Latin , he's well-read . He knows history , he knows art , he knows religion . Where are those people going to come from in the future ? If we just have a bunch of skimmers , we might as well let the whole world be run by AI .

Yes , we have Daniele , and polymaths and people who are exceptional .

Yeah , if you want to say , give me the first cut of what the generative AI tells me I should do for my term paper , you know you're going to get a mediocre result , but the way AI works is you can continue to interrogate the AI and be able to get access to things that you might not have even thought were ever there , even if you were the most deciduous

researcher ever . The question that I have is along that line if we are going to allow the algorithms to basically make the decisions and those algorithms are making decisions within a range of guardrails right of accuracy on one end or another .

So if you were in a company , for instance , right , and let's say you're an insurance company which you happen to be quite familiar with , that as a chairman of a large insurance company very risk-oriented , if the algorithm says you're going to be good if you're between this percentile and this percentile of risk , if you are an orthogonal thinker , an original thinker

, and you want the organization to take a risk that is outside of those guardrails , doesn't AI kind of preclude that ?

Only thing you need and I do think you need this you need a partner like him at the top , looking at it from a call it a human standpoint and bringing in these other outside influences that maybe didn't get baked into the machine's work . So what does that mean for Deloitte ? Well , that means they don't need a lot of staff .

They'll probably take the position that well , the fee stays the same or goes up because it's a value-added fee . It's not based on rates and hours , but it wipes out the need for staff . So what you need are beginners to do various day-to-day activities . You don't need anybody in the middle and you need partners .

Well , how do you become a partner in a big firm like Deloitte ? Well , you can't go from a beginner to a partner . There's got to be some kind of bridge there , and I'm not smart enough to understand in that profession what the bridge is . But I can see that the world of external auditing is going to change dramatically .

Economically it'd probably be better for those that survive , but it's going to be a lot smaller firm .

Whether you agree with Marx or not is irrelevant , but the fact that he changed society in a massive way is indisputable .

And with enough rhetoric and repetition and argumentation , they're able to bring large groups of people to see the world in their way and say and therefore , this probably is the way of evolution . I mean this from a scientist perspective . This is like Thomas Kuhn . And how does scientific discovery actually occur ? It's not an aha moment .

With E becoming MC squared , there were decades of work leading to E being understood and MC squared being understood before the two got put together . So there is this evolutionary feature to the way in which we evolve and therefore you need these broad spectrum thinkers who are on the edges and fringes of thinking to actually cause change to occur .

With a specific case example , the one that is cited in the book is about what happened in Myanmar , where algorithmic biases inside of meta's algorithms resulted in cornering sentiments of an entire nation to the point that a fringe element in that society was brought to great harm as a result of algorithmic bias .

You know it goes to your earlier point on well , is there a legal reason now to find fault with something ? Those are untested waters .

But this is happening as we speak because engagement is leading to an echo chamber effect , where people are self-sorting and saying I'll listen to this and not to that , and the more you listen to it , the more you're convinced that this is the truth and now you're willing to take action on the basis of that , by the way , for the benefit of people who haven't

read the book or are not necessarily familiar with it .

Otherwise we have the potential to let this start to tell us what to do , and that is the wrong thing , because , as I said , this is nothing more than a token prediction machine . It's not intelligence . I think we are attributing the wrong values to this by misnaming it as artificial intelligence .

We moved to generative AI , the next thing is agentic AI , and then we moved to artificial general intelligence and then artificial super intelligence . Well , you get out to that extreme and you'd say , well , that's way beyond my lifetime . Well , it depends upon what your age is . It's coming faster than any of us had predicted . Well , certainly for our kids .

They have a combination of intelligence , super intelligence , of knowing everything ever known in the entire universe , as well as a level of self-awareness and consciousness , and I think that's where we kind of that's where it doesn't work , from what the reality is , of what artificial intelligence is . There is a big difference between intelligence and consciousness .

You needed to have a good combination of all three and have a balance of all three in order to reflect the fact that you are actually human . Well , I think , with these intelligence machines , logos is going to get way past our capability .

The average human will not be able to keep up with the logical inferencing that these machines have the capability for , because they can take on incredible amounts of data , connect the points and make good decisions and much quicker , much quicker and make them better , faster , cheaper than we humans are able to . But will it replace my ethos ?

Unlikely , unless I teach it the very meaning of ethos , which is very difficult . We all know that codifying ethos is the problem of the legal body .

But when you get to pathos , there is no machine in the world that can replicate my pathos . So I think there are certain boundaries that are self-inflicted here . We just need to be recognizing of that and say lean into those , because those are the things that will not change just because we have better machines , ashwin .

Ideas are going to become more important , not less , because the tendency is going to be to be mechanical , to do things by road , and what we have to think about is the real response is to actually accentuate these classical ideas pathos and ethos in particular , but there are certainly others as well and I think that's one of the problems that we in the education

business , if you want to put ideas here pathos and ethos in particular , but there are certainly others as well and I think that's one of the problems that we in the education business , if you want to put it that way , have to deal with , because our kids aren't being taught that well , and that's actually what .

at universities me in particular as professor at a business college what should , what do you think , what are your recommendations on what we should be teaching people kids , mba people , undergraduate people what kinds of things should we be sensitizing them to ?

Maybe the corporate boardroom may be too much of a sophisticated environment that they're not going to initially have a pathos for , but what do you think we , what kinds of things we should we be getting across then ?

When I joined boards two decades ago , in the morning we would have a strategy discussion , in the afternoon we'd have a technology discussion and then after that it was regulatory issues . That no longer is the case . The technology and the strategy discussions are the same .

So you can't be an effective board member without having more than a passive knowledge of what's happening in the technology world . Again , I'm not suggesting that everybody needs to be a technology expert , but they have to be more than a casual reader of the back page of the Wall Street Journal . Now , how does that translate to people being educated ?

I think whoever it is , irrespective of what their career aspiration is , they have to have a technical knowledge . You'd need to define what that really means . But it has to be more than just a casual relationship with technology , because it impacts business . It impacts certainly the board and management .

It impacts certainly the board and management , but it impacts all of us in a daily manner . So I think what you teach the younger people there has to be a large segment of technology education and knowledge .

There's a lot of mechanical considerations that we should be teaching our children because to survive in this new world , they need to have those mechanical skills . You know , data , data , analytics , understanding that , python , programming . There's a whole bunch of things that they should be taught .

But in addition to that , I think there are certain values that I would gravitate towards as sort of timeless values . It'll take a lot of wisdom to make good calls in the future , otherwise you're likely to succumb to a machine making the call . We need to train our children to understand the meaning true meaning of justice .

Justice is oftentimes in the eye of the observer , but I think there is a higher call which says that there is just and unjust , and you need to distance yourself and look at it from a higher elevation point and see whether it is just to all concerned .

I , for one , particularly feel that because I grew up in just immediately post-apartheid Africa , so I have seen firsthand what the lack of justice can do to society . It can really harm generations in society . It takes generations to recoup from those losses .

It also says that we need to teach our children courage , to speak up when they're convinced that they're in the right , not just follow the herd and the mass mentality , but to stand up and be counted and , last but not least , to be moderate about it .

Otherwise , you know , you could be wise , you could think that you are being just , you may be courageous , and without moderation you become an extremist . So we need to throttle that instinct and say you may be right , but you still , and speak when you have to speak , take a stand only when you must Be aware that there are countervailing forces at all times .

I think these are things that we're not teaching enough , and I think that goes to the classical foundations that we need to lay , and I think we need to double down on that , even more so now .

You know , I learn by being in a family that there is times when an elderly person is in the house , I have to do certain things , I don't say certain things . Um , you know , if I have a someone with special needs , I've got to deal with them in a different way .

If those kind of human experiences which have been stripped away are gone , then I think that AI can be very dangerous , I you know , because these are people who don't have human values .

I mean , I , you know sometimes that , you know , you talk to some of these tech guys and you're just saying , really , I'm going to , I'm going to , I'm going to entrust the future of humanity to you .

Bucking the trend has the potential of being leaders if you can convince other people to follow you . We've been questioning the value proposition of higher education , and the whole tenor of higher education in the last several decades has been to create specialists around particular , Very narrow , sometimes , Pieces of knowledge . Right .

I think of Sorokin's remark that we know know more and more about less and less well , and what we're really saying here is that we need to reinvigorate the university with the same kind of values that basically started us down the road in ancient Greece , right , and teach them logos , pathos and ethos and have exciting classrooms where- .

Well , yeah , and apply it in a modern metier right , In a modern- .

That's something that I think the universities have missed in this rush to specialization .

Yes , a lot of things that couldn't be done previously can be done in the future . I mean , I think of the fifth generation of AI as artificial general intelligence . The sixth generation of AI is where robots are infused with artificial general intelligence , so things that were impossible are going to become possible in the near future . Should you do ?

It is a value-based judgment and I think we need to teach our kids how to have that discussion with courage .