Navigating the Workplace in the Age of AI: Hear from Industry Experts and over 500 CEOs on How AI is Reshaping the Workforce of Tomorrow

Thank you much, and it's amazing to be here once again and partnering with Achieve Engagement, and of course, the edX slash two team, , because we have an unbelievably informative and educational session ahead of us today My name is again, Dan Shaw Bell I am the managing partner of Workplace Intelligence, three books, , over 70 research studies, including the one we're really focusing on today And I have a distinct pleasure to be joined by two highly intellectual, , men Anon Al, chief Platform Officer of two u, founder of edX, and professor of electrical engineering computer science at m i t as well as Korean, Ani, the Dorothy and Michael Hints, professor of Business Administration at the Harvard Business School , just a little bit about them.

They have very extensive backgrounds I'm sure, , you can Google to find out even more about them We're gonna be obviously leveraging their expertise as it relates to this topic around AI today Anon has previously served as a director of C A l, MIT's, computer Science and Artificial Intelligence Laboratory, and Kari's Digital Transformational Research investigates the role of analytics and AI and reshaping business and operating models He co-authored competing in the Age of ai, an award-winning book published by the Harvard Business Review Press.

He's also developed six online courses that have educated thousands of executives on AI strategy, technology-driven transformation, and entrepreneurship , as I said today, our focus is definitely on navigating the workplace and the age of ai Obviously, most people are either interested in ai, who are, who are, who are calling in from all around the world, are currently using it, or trying to implement it in the organization, using it for collaboration to, , , speed up processes and, and, and performance , and both Anant and Korean bring incredible amount of, of breadth and depth and credibility to this topic as it relates to education, , corporate workplaces and through a technology lens as well And my company, as I was saying, is workplace intelligence.

We partnered with Anand's team at edX They were an amazing team to work with on this, to conduct a new study of 1600 full-time US employees and 800, , 800 of which were the C-suite And just over 500 of the C-suite were actual CEOs it's an unbelievable sample as well as, , employees that were sampled and a non Corin will share their perspectives on the study and explain the implication to your organization as well to upskill and prepare your workforce for the future of ai And guess what the future is now.

we need to start really learning and educating ourselves on this topic It's, , on the utmost importance , as, as we said before, be sure to lead your question in the q and a box we can make sure we can answer them at the end And without further ado and not encourage, it's great to have you here, Jennifer, to be keep follow to the conversation Awesome.

Well, we're gonna start with Kari, based on the research idea that we did, why do you think there's a disconnect between employees and the C-Suite executives on the magnitude of how AI is disrupting the workplace Workplace? Yeah, I think, I think this is, , this is a great question And again, it's great to be with Anant, , who is a amazing leader of what he's built with edX , and I've been lucky to have one of my courses on edX , and the, the amazing, , impact of having worldwide for education is, is incredible , what I would sort of say is that there's a really, a knowing and doing gap about ai, , , in my, , interactions around with many thousands of executives, they all say , oh yeah, it's important.

It's gonna change the world It's gonna change my life It's gonna change the jobs lava But very few are actually actively either engaging in projects or actively doing the work that's needed I do this little test these days around generative ai.

I ask these three questions I ask executives, how many of you have tried generative ai? Most people 9% have tried Chad G P T Then I said, how many of you think that in three years time, in three years time, it's gonna change your job, your company, your organization, your workforce, 90% stay up and say yes And I say, how many of you are using generative AI every day today? Only 10% say yes There's a knowing doing gap, ? They say it's gonna change their world in three years time, and only 10% very smart people in 10% of the time are saying, , that I'm actually using it every day.

And this knowing doing gap is because people have not invested in the learning It's kinda learning how to ride a bike, ? when you're 40 and you're learn how to ride a bike, very different than when you were eight And if you're 40 and learn how to ride a bike, ? It's gonna be painful You're gonna fall, you're gonna embarrass yourself People will laugh at you, but you gotta do that work to be able to become an expert, Dr.

A biker But then once you get it, it's natural You can't unlearn it I think this is what's happening with ai There's a knowing doing gap.

There's a lot of worries about it, a lot of fears about it, but that fear is actually coming from the fact that people have not invested in the learning , Kareem, , I just love the analogy, although when you said 40 years, I would say more 60 years, but, , yeah, yeah, 30 plus 35 plus 35 plus everybody 35 plus But, but, , but that said, I loved your, , , bicycle analogy, Kareem , what, I've been advising leaders, whether it's CEOs of companies or presidents and chancellors and, , , , professors and on at universities I said, , I said, just today, you have a window, usually open the Google search.

Pretty much all of us have a Google search window open at the ready We also have Google search on the homepage of our mobile phones Then I ask them, how many of you have chat G P T on your home screen? And in big service that I've done informally at conferences and on, the answer is zero Yes And I tell them, just you have a Google search window open, you should have a chat G p t or equivalent binging or barred search window with gen AI next to it.

we begin to develop the facility to play with it and, and work with it And I just totally agree with you that it should become a bicy Riding a bicycle should become natural to you And by learning that, that's when you begin to start using it And, and the same thing applies not just to CEOs and university presidents, but frankly, to everybody in the world, whether you are a frontline employee, executive professor, doesn't matter.

You gotta do this Yeah , , as you say that, talking about universities, I was, I was an elite Ivy League University, not Harvard recently of ours speaking to their, , their cabinet, , basically the president, vice provost, , the provost and the deans and their executive And they also said, this is really important, but none of them had done the investment in, in, in learning Exactly.

And they're in a learning institution That's the biggest puzzle for me You're , in a learning institution, this is what you gotta do , , , and and , , I think, again, this, this, it's, it's obvious what needs to happen Absolutely.

Yeah And the, these stats from the study, as you both, you both alluded to really showcase this gap, ? , 59% of employees are currently using AI at work, and 85% believe they'll be using it within the next five years , why are they waiting for five years? And the other thing that's really important to note, which kind of alludes to what you were saying too, is, , you could pull up chat G B T for instance now, and start to just play with some prompts , it doesn't take too much time That's the whole point of it, ? And it doesn't cost you anything.

we're gonna, we're gonna move on to this, this next section, really about the C-suite This is, in my opinion, perhaps the biggest, most interesting finding , there's been much talk about, , the workforce being more automated and, and the workforce kind of fearing AI and what impact it could have for their jobs , but now even the CEOs admit that their jobs could be automated or, or affected by AI in some regards What do you both make at this? what, maybe I'll jump in first, Kareem, , when I first, that statistic that 47% of C-suite executives, including CEOs, believe that most, if not all of a CEO's role could be, , , replaced by, , ai.

I was, I was frankly stunned I did not believe it Because, , generally you, you, you believe that as you become a manager and then a C-Suite executive and a C E O, your soft skills and, and your human skills, , critical thinking become more and more important at things that are harder and harder for AI to do , I was very surprised And then as, as I talked more about it, I realized that it actually made sense.

And the reason is that it, the C-suite executives and, and others at, at the high level are taking executive education courses on AI for business , they are talking to other colleagues and on and all hearing about the power of ai They don't have their heads stuck in the sand , they're not using gen AI yet, but they know that it is gonna be transformative And even more from a technical standpoint.

The reason I, too now begin to believe that is if you look at a CEO's role, particularly A C E O, who does not have a great, , does not have a great, , , EA or a chief of staff, I assume they don't have a great chief of staff, 80 to 90% of the role And I can say that from personal experience as a C E O who did not have a chief of staff, 80 to 90% of your role can be automated You spend a lot of time writing emails and, and saying, , yeah, you agree to speak, or you, , or, or emails saying, oh, you cannot, , make the event , you, , write reports for your board and for the others , you're sending emails and discussion points to partners preparing presentations.

, a lot of that, only about 10 to 20% of a job is truly parts where you are inspiring your employees, hiring great people, , , getting the word out, thinking about dreaming about the future, , dreaming about new products That's, those are kind of things that CEOs should be doing, but only 10 to 15% of time really gets spent on that I think what's gonna happen is that AI is gonna automate and do all the things that great chief of staff would be able to do And I think in that sense, CEOs will become much more empowered in the future What about you? Current? I totally agree with you.

What's your take this? I, I totally, I totally agree with you And I think, look, I think, I think the, the, the key here is that the architecture of, of work is now gonna change, ? The architecture of how we get work done is gonna change If you sort of look at something that Anant is really familiar with, , I was also, as an undergrad, I did, I did electrical engineering, ? , and how important software is for electrical engineering, for electrical engineers And that has really, , , we, we were talking about ways in which our circuit design education has changed radically because you brought this tool in to massively improve the productivity and quality of electrical engineers to do very complex circuit design Well, those same tools that were invented 20, 30 years ago in this very narrow field, the circuit design, because of complexity, are now basically seeping into how you manage your work, how you manage your employees, how each, each knowledge worker manages themselves.

And there was a learning curve, ? Or not, in how people had to go adapt, adopt these tools And I think the same thing is, is is now happening across the board in every occupation along the way Absolutely And what's interesting about what's, I was just gonna say, what's interesting about STAT is that there is the disconnect that that builds off of the last thing we talked about Only 20% of workers believe most or all of their role will be replaced by ai.

CEOs seem to be more on the know, in terms of the transformational nature of AI as it relates to their role compared to workers , I think, look, I think, and, and again, annun sees a lot of, a lot of CEOs and forth , what I would sort of say is that, look, , I think, I think that the real challenge, look, about a hundred years ago, a Alfred p Sloan basically set the stage for the modern American capitalist enterprise , and that that model has been copied, , everywhere in America, , the General Motors of the world, the GEs of the world, the IBM's the world, follow that model , I think there's a time for a new model.

, and I think part of what the anxiety that people are feeling today is this anxiety around what is this new model gonna be? How do we train ourselves? How do we keep learning? How do we actually change, , the structure of our work? I think that's why the future of work is not just about remote work or not work, or remote learning and learning It's really this actual architecture of the interactions between employees and machines, , and the hierarchy I think that's where the exciting things are going on , I think it's, , Mac Kareem, totally agree , I, I think in addition to between employees and, and, , , leaders within the company, I think, I think the, the place where the biggest impact is gonna be is between your customer and you as a business where AI is gonna take up a large part of that role in creating a much closer contact with you, with your customer.

, and by the way, , , at edX, and, , we, we talk about this, but then we eat our own dog food here if you go to edx.org on the homepage, , , as you click through the various courses and on, in the homepage itself, you will see a, , edX expert edX expert is our AI based, , suite of tools that work with learners and consumers, , in all kinds of ways Contextually, if you're on the homepage of edX, it'll help you find new courses, , , go, go to edx.org and play with it , and, and what, what I'm hearing learners tell us is that previously they had a hard question, , they would have to wait hours or sometimes for a longer time for a human to, , , respond to them.

But with ai, they get responded to instantly, and they feel much more comfortable engaging with AI because, , if you're asking a lot of silly questions, you feel a little embarrassed to ask those questions of a human being But if it's, Hey, this is only ai, it's a machine, I can keep bombarding the machine with silly questions And we've seen the same thing We've launched, , an AI based tutor in our courses, in my own circuit scores from M I T and edX We've launched a tutor, , bootcamps have launched tutors on, , edX.

And we are finding that learners are six times more likely and more interested in engaging with the AI tutor than with the human tutor And the reason they're giving is that they feel much more comfortable asking silly questions and bothering AI at all hours of the day or night, and not worrying about hurting AI's feelings I love that And it's agree , yeah, I would much rather ask what I think would be silly questions to ai, rather than, , pull my manager or talk to a customer about those things to get educated before I engage, ? I think that that's very, very applicable.

And I think your point about the changing nature of, , company, the company relationship with customer and ai, improving that relationship in, in those, in those conversations and what's focused on, I think that's gonna be a huge change , great points And I, I think of course, with both of you here, , focusing on, , skills and learning and, and, , , workplace preparation I think is a really, really important topic And the question is, , are employees ready? Can , can these companies find the talent? , because we found that the C-Suite executives interviewed or having trouble finding this AI talent, or those individuals with the AI skills to be able to be efficient and effective at their job, , even though they're willing to pay them more, that the pay doesn't seem to be a big enough incentive for people to learn these skills yet , and why is it difficult for companies to find these workers, ? But Dan, , , the, , our workplace intelligence survey, , with you, , shows that, , 87%, , 87% of C-suite executives are saying that they are having trouble finding, , , , AI talent.

, and it, it's not just talent specialized in ai, but people who, who understand how to use AI in their work And even more, , importantly, probably is that most executives also believe that they are, they are perfectly willing to pay people who understand and are skilled in ai , 82% of them are saying that they're willing to pay the workers more not only are they finding it hard to find skilled workers in ai, but 82% of them are saying that they're willing to pay them, , willing to pay them more given that, , , it should be a, , , a, a a a call to action for everybody, , to get those AI skills.

And it's actually not hard to get those AI skills , there's more and more, , , free courses available , and of course, you can just go and play with chat G p T and become a quick expert yourself We've launched a number of courses and programs in prompt engineering, , , where it, it tells you how to interact with, , a gen AI in your job, , , in a, in a broad set of contexts, and you can learn about it in a few hours if you go to our site and you Google for, , , a prompt engineering, it'll help connect you to something.

And in the same survey that we did, , one of the big conclusions that jumped out at me was that the C-suite executives say that the single most important skill that you can go and learn at this time is prompt engineering When prompt engineering is becoming a whole new field , just in the past, , you did mechanical engineering, civil engineering, electrical engineering, prompt engineering is a engineering discipline that teaches you how to work and engage with ai And, and, and that skill is probably the single most important skill for this 21st century And you're saying that because you really need to know what to ask the machine, ? In order to get what you need? Exactly.

But you need it It's not just, but Google search, you type in a phrase and it shows you a bunch of stuff It shows you pages and pages and pages, oftentimes hundreds of pages of stuff Then I have to go through all those pages to figure out what I need But with ai, you get one answer back.

your question better be good, otherwise, the one answer that you get back could be hallucination for all that, And , prompt engineering is the art and science of how do you engage with ai on the on, in the chat, , one of our listeners, Craig pointed out that, , you got to, , provide guardrails part of prompt engineering is not only to ask a question of ai, but provide guardrails things you can tell ai, look, pretend you are a professor who understands, , , AI or pretends or, or who, someone who understands, , civil engineering and, , , answer my question, but here are the boundaries of the answer.

, , prompt engineering helps you constrain how AI provides you with answers Yeah If I may add to this, this is interesting, Anan, in many ways, anan, my view is that the, , for close to, , since the web browser was invented and search was invented because we had all this information available to us, , what the web browser did, is that it lowered the cost of information dissemination, ? Basically, we, we went the marginal cost of information to some went to zero Everybody could now put on information on the web and be globally available And then search became a bottleneck, search became a bottleneck, and then we learned to search, and the search engines got better and better over time.

Remember, Yahoo was , then there was Alta Vista, there was excite, all those search engines were out there, ? , and then finally, Google figured it out really well using graph theory , , and then what we learned is that we learned to get the answer on the first shot with Google And the answer is the webpage that's the metaphor that we have trained ourselves over the last 30 years to, to search on the web Well, cad, G P T and Generat AI systems are not a search engine.

People are misusing them as search engines They are cognitive partners They think with you, it's a genius intern that knows everything, but can make mistakes And your job is to use it as a cognitive partner to make it work for you And this distinction is actually really important because if you use it as search, it's not gonna work, it's gonna fail.

And then people say, it doesn't work for me Well, because you're using it the wrong way, ? It's , it is , it's trying to ride a bike in a swimming pool It's not gonna work for you It's a different, it's a different tool for different use case And the use case is a cognitive power, ? Lowering the cost of cognition.

And if you sort of imagine that, and there's people sort of, sort of saying that, is that it's lowering the cost of cognition, then how we learn to use it is very different And that's why courses prompt Engineering, investing in the learning and knowing where it goes wrong, it's gonna be important I'm gonna put in the chat, a study we just completed with the Boston Consulting Group is an experiment We've tried to see our chat, G P T will impact, , consulting, ? High-end consulting work, and we have good news and bad news The good news is that much of what our consultants are doing can actually be done by Chad, G P T, and they get better, ? And , the things I say is that machines won't replace humans, but humans with machines will replace humans without machines.

That's the old adage in computer science that has come back really importantly, but also because of these systems are weird, and we're just now understanding them because ai, generative AI is good at creativity, but bad at math, we never imagined that AI could be bad at math It's based on mathematical principles, but it's bad at arithmetic, ? And then you have to know what the limits are and figure out what those things are at But thinking of this as a genius intern that is side by side, and it's your co-pilot, I love Microsoft's analogy of a co-pilot It's a co-pilot, but you're the captain You just have to take responsibility for it.

And don't get mad at it You haven't asked the things You haven't actually, you don't, getting mad at your interns too often is bad, ? Because they'll quit Now, the nice thing is that G P T will not quit on you It has, , infinite patience.

, Kareem, Kareem wait till, , , wait till we start imparting emotion to ai But, but at least, oh, I know, I know Oh my God , , at least that happened Hasn't happened yet.

my AI does not have an attitude But can you imagine, once AI gets an attitude, then all hell will break loose But at least until then, let's, let's make hay Exactly , I, I, I love your analogy of a copilot, , Kareem, , Microsoft and GitHub have launched, , , a a, , their Python programming environment has, , a co-pilot built into it.

Yes , I, I to, , and I think every one of us can have used Chad g PT as a co-pilot to help us by having built on our screen , , I liken it to, you can call it copilot, , having been a c e O myself, , I liken it to a chief of staff where yes, a great chief I as well, a great chief of staff can make a c e O much more effective, , because a C E O can spend their time on things that they believe are super important meeting with partners, meeting with human beings, meeting with employees, meeting with customers Those are the important things.

And , , we imagine every one of us can now have a chief of staff , this is, this is, , , I, I tell you folks, this is game changing Wow Those were great examples and, and very well explained, making it really tangible for everyone , what ba from both your perspectives, based on everything that you discussed and all the, the stats and findings that were shared, if you were a c c-level executive, or in, in this case, a lot, we have a lot of HR leaders as attendees, what would you be doing now? And what advice would you also be giving to employees, especially with the need to kind of enhance their resume and, , skillset focused on ai? , maybe I can take the first crack at that, Dan.

The same, the same workplace intelligence study, , gave us a, , a sort of very, very interesting and, , conflicting set of numbers What we found was that, , about 60% of employees in companies were learning, or wanted to learn more about ai But only 20, about 20 to 25% of employees said that they are finding these opportunities in development from the HR departments and companies , look at the gap 50 to 60% of employees said they wanted to learn more about ai, and only 20 to 25% said that they were getting these opportunities.

, a two x gap why don't we call this the AI learning gap? , frankly, my advice to folks in hr, or if you're a leader in a company or a C-suite executive in a company, , I would look to immediately find opportunities, , to give learning opportunities in AI for your employees Else It's simple They're gonna leave, , , 87% of CEOs believe that, , , people who know AI should be paid more, and they're having trouble finding people.

I think there's a huge, huge, , , set of, , , events happening now , the perfect storm where there's incredible opportunity for people they wanna learn And, and the ways to learn is pretty simple , , there's much available online and around you where you can learn, , as an example, , for myself, , we've launched AI for, , edX for Business, where we offer a catalog for executives at companies to offer their employees, , where they can learn about all things AI from, , , all, pretty much all topics you can think about, , machine learning and ai, how to apply AI in various disciplines, AI for executives, you have executive executive education courses We've also launched a master's degree with UT Austin, a top 10 ranked school in AI for just $10,000.

We've also launched a bootcamp in ai, , where you can learn about how to apply AI in, , developing software And the demand has been, has been absolutely incredible , all of these learning opportunities exist where you can learn for yourself or in a company, you can offer these opportunities for your employees no matter how you do it, I would say there should be a rush to offer these opportunities for your employees Otherwise there's a real danger that they'll go off and, and get these elsewhere and, and move to those companies.

Yeah Another, if I may add to that perspective, I think it's a very excellent perspective I would say, , I, I think there are, there are, for the HR talents managers that are here, I think there are two big, , messages that I would give them First is, when we have studied the deployment of AI inside of existing companies, what we see clearly clear clearly that it's 30% technology, 70% organization, 70% organization is all about org change It's all about culture change.

And that's where HR has to play a really important role 'cause guess what? Your technologists are terrible with people, ? That's why they became technologies That's why Anant and I became technologists We were , we didn't working with people, we went to more computers, and now we realize that we gotta talk to people And for sure, you have to basically in, if you, if you, if your company's serious about AI transformation and digital transformation, 70% of the work is cultural, organizational, and change.

And HR leaders have to be the partners, both with their technology leaders as well as their C-suite to make this happen First, first, first thing The second thing I would say is, I actually hate the term learning and development L and d, L and d learning development is the wrong thing Because what I keep seeing there is that it's a checkbox, oh, did the course, did the course, did the course, it has to be learning and doing, still keep l and d, but change the development to doing all these AI educational initiatives.

It's not enough to educate You gotta translate into doing, taking action the learning and doing actually has to be tied to actually use cases that get developed by these folks Don't just check the box Lemme give you an example.

, , I've been, I'm pretty close with the folks at Moderna I've seen them, I've worked with them, , I've done some advisory work with them and forth And the C E o, Stefan Benell has made 20 hours of AI training mandatory for every employee, every employee, no matter what level you're at, everybody has to get experience in thinking about ai But it's not just the, the taking the coursework, which they developed with Carnegie Mellon They actually have to then develop, use cases for their functions, for their roles as part of this thing.

This is the doing part I am fully behind learning, ? That's my job , that's what I get paid the big bucks for But the more important thing for the HR professionals is to shift from development to doing You gotta become the doing partners with your technologies, organizations, with your, , with your functional executives to make this happen.

Excellent Well, I think that's great advice We have a ton of questions as you both can see , the first one I think is really interesting , how do you advise organizations to navigate the legal issues around IP if using generative ai? Or do you just advise on an analytical tool? You have a perspective on this, ? Yeah, please go ahead.

No, no, go ahead You go ahead No, I was gonna just say that, look, we are at the early stages of this technology, and there's gonna be a decade's worth of litigation around ip Most people, what they're worried about when they talk about IP is , if I put some proprietary data into chat, G P T, is that gonna then leak out to the rest of the world? Well, the answer is that all of these systems, ? If you use that G P T or, or, or ANTHROPICS code or whatever, Google releases all those models, ? , if you use them through APIs, then those, , have guarantees for, for data safety and data privacy First thing in the consumer versions, ? For chat, G P T, , for sure there, if you get the pro version, if you're paying 20 bucks a month for it, then basically in that situation, you can have private browsing where you can upload whatever you want and it won't train the rest of the models.

those things are already built in, ? The broader IP consideration about the fact that they've learned off the internet, they've learned off many, many content bases Is that, is that copyright violation or is it fair use that's gonna be litigated over the next decade? , I think Kareem well put, , , I'll just add briefly that, , , we are using, , various AI tools ourselves We are working with both, , , Google and, , OpenAI using their tools , for instance, , as we use the enterprise APIs in these tools, , , both, , Google and the chat gp, , OpenAI make pretty clear that any information, , that we send over the enterprise, a p i endpoints is private information where it doesn't, , leak into training the global model They're very clear about that.

However, , we've also released a plugin, , for chat, G P T, and, , if at a plus user, , the, you get access to the edX plugin there, it's a little more nuanced there Any data we send open ai, , is, does not train the model directly, but any answer that chat, G p t synthesizes from the data we send them can be used to train the model these are being worked out, and there's a lot of nuances and, , it's very early days And I think these are kind of things that we need to continue working with the big providers and make sure that the end consumer and humanity at large is protected I really this question is stating that AI can replace the CEO's role.

The same saying that AI can fully run a business, start a business if given a prompt in the future Well, listen, you don't wanna take that I, I listen, , look, I read a lot of science fiction In fact, that's the best way to understand the various features that might unfold for us And there are certainly scenarios where, , , machines do a lot of things.

And, and certainly there have been attempts at those things before , T V D I, I don't think, not yet today, but will there be attempts at this to automate many, many things? Of course, that's gonna happen, for sure , I, I, , I'm sure many, , many folks here would be super excited, , to have AI replace some of the top posts and, , and democratize, , democratize top positions and salaries But yeah, I, I, I wouldn't, yeah, I, I wouldn't rush to replace my C-suite executive or c e o because remember, , just because AI can do 80% or 85% of the work that A C E O does, it still says that the remaining 10 to 15% is, , very unique and critical And , until, , I'm confident that AI can do a hundred percent of the work, it'll be really hard to, , hard to do this until then, , I think using AI as a chief of staff or a co-pilot, , is the way to go.

Definitely One of the top, another question, one of the top criticisms from the C-suite level naysayers is that it takes a lot of compute to train or fine tune AI systems With the amount of new data getting generated every minute, this cost of learning is going to go up and higher How would you feel or respond to it? Maybe I can start taking a crack at that , , but certainly at this point, , , there a lot of compute and, , , , the, the stock of NVIDIA and a M d, , and some of the other silicon manufacturers skyrocketing is testament to the fact that, , chips that support or accelerate the use of ai, , are in huge demand.

And certainly at this time, it is the case that, , that AI's responses and on are very unsustainable , these big data centers, , are one of the world's most non-sustainable places burning a lot of heat and, and on But that said, these are early days , it's not at all unusual to find that computation is expensive, and AI algorithms are, , not, , they're very expensive algorithms, but a lot of scientists at university and companies are already working, , aggressively on finding much more efficient approaches , , architects are working on much better architectures for making AI calculations become more energy efficient and, and simply faster.

, , again, I won't, , , I'm, I'm not too worried I, I think within a few months, two years, , we will have, , , we will have better algorithms and also caching technologies Where today, even at edX, , if you want an answer to something, we send a whole lot of data, but caching technology where you can cache responses and you can, when questions are repeated, you can use the cached information to respond to a user versus asking Chad GP the same question again and again and again , just caching as one technique will, , reduce the amount of computation we need, , , in huge measures Yeah, absolutely.

And more generally, what I was gonna say is that, , the history of technology always points to this, ? imagine the a hundred years ago, the automobile was invented in Detroit, and , in Stuttgart and Munich, ? In Germany, , and, , the Ford model t engine in from the 1910s, 1920s, ? Or the Mercedes-Benz engines at that time, , were massive only produced, , five horsepower or , ? And we're highly polluting, ? And if you did straight line projections of performance, ? You would need a, a what the performance needs to be today for engines, ? You would say, well, then that means that, , we'll need a, a car engine the size of Detroit, but what happened over the last a hundred years, ? Engineers get got at it, made it more efficient, made it more powerful, made it less polluting, ? And we, we can't think about this technology in cross-section today Yes, today there the compute costs are high, the data costs are high as well But that's a transient thing, we're it has literally just barely crept out of the lag, ? And now there's gonna be many startups, many large companies that are trying to figure this out because they're faced with economic constraints Computers too high Well, we're gonna throw engineering at that.

Data requirements are too high Let's throw engineering at that And all those things in the end, and I'm totally with an odd, all those things are gonna make a huge difference How can we design training programs that empower employees to acquire the skills needed to collaborate effectively with AI technologies and adapt to evolving job requirements? Sorry, is that a question or is that a, a statement? That's a question Yes.

The answer is yes , I, I think, again, , these are very early days and, , I think, I think what OpenAI and chat G P T did was suddenly created a quantum improvement in the technology, particularly for kind of the cog cognition based AI responses And, , brought it to the mainstream, made it very easy to use And what we are seeing now is a huge host of projects that people have undertaken in terms of what is called blended ai , how do or co-pilots, how do humans work besides ai? And your question about how do you bring AI into teams where, , you work with colleagues and work with AI and, , solve problems together.

I think that's gonna be a fantastically important area And Kareem, maybe your next step, a next Harvard Business School study and report will be about how to do that Hey, maybe K we should launch and launch a course on edX, on how to work in teams with ai , for one thing, , AI's feelings will not be hurt if I say stupid if I say something, but I, , but I'm al but I, I'm, I'm always very kind to my robots, ? Because you always say, please, and thank you, and make nice requests, because I'm also, I'm also a believer in Pascal's wager, and who knows when the singularity comes that all the data that has on them, they'll know if you were nice to your robots or not nice to your robots I'm always , what? Let's just be nice to, let's, it's always, , my mom always taught me, my grandmother always taught, my Nima always taught me, , , be nice to people, also be nice to your robots.

It, it doesn't cost too much What's really interesting, over the past, maybe year or , there's been talk, , , diversity, equity, inclusion, belonging has been a thing for a very long time But from the diversity standpoint, robots are now part of the mix, ? Yes They, they are now part of a team , when people think about their workforce, , maybe they have a thousand humans, but they're probably, as you're saying, with, with, , everyone having their own, , co-pilot.

Maybe there's 2000 workers overall, ? Yeah , , one of the great, , and this, , Eugene, we, , who's a blogger and a technologist, wrote this amazing, amazing article about TikTok and the algorithms of TikTok And what he said about TikTok, which is, which was cool, was that at TikTok, they've been successful with the algorithms and what they do, , what he said is that the algorithm has a seat at the table in the product development, product design, product execution, , , conversations the robot will soon have a seat at the table It has a seat at the table.

Now, with me, my, my my chat, g p t robot, my Claude robot, my, my, all those things ha is always on I'm always consulting my robots for advice and for doing work and that kind of stuff, for crafting new emails, for looking at suggestions and on and forth And , , and , then, then basically, , we wanna, we wanna sort of start thinking about them as having the seat at the table and also being a cognitive partner , Kareem totally agree Yeah.

There's a nice little thread, , running on the webinar, on the chat, , , , with, , Irene and Sesar and some of the others that talk about how, , , politeness, , is a learned skill Yes And We learned politeness from our parents And, , certainly as I was growing up, , , my dad was, , , my mother was very patient, but my dad was extraordinarily impatient And, , , if, if he would ask him a question, , , it is not a great thing.

But he would, , , he would respond very grumbly And , , , but with ai, as AI continually responds politely and keeps answering questions as our kids work with ai, I think it, it'll teach our kids patience And we've seen that happen before , if you take Uber, for instance, , I think Uber and Lyft have taught our kids patience and politeness because of the Uber ratings , if you're impolite to a drive, , , Uber driver, , then you get a bad rating.

Yes And people learn to be polite, people learn to be nice And, and I think that is training our whole generation to be nice Don't slam the door in the car, , say thank you after the ride And I think it's creating a whole new generation of people that are nicer.

And , to me, this is one of, this has been one of the great victories of technology Yeah I'm a, I'm a 4.7 on Uber, and I, I'm not sure why I think I might have, , been too tired one day, and I, I'll say hello or goodbye, , but I a hundred percent agree I think that that's a, that's a, that's a great perspective and a very tangible example.

Yeah, That's a, there's, there's a great, there's a great bit of bit of chat also about the downsides of ai, and we don't want to, we don't want to actually ignore that Yes, there, there are some amazing downsides of ai The way I think about that is that, , AI systems grow exponentially in value, but you can also exponentially grow the bias You can also exponentially grow the downside we actually have to be super concerned about that.

There's worries about how representative the, the data is that it's being trained, how bias, if it's training off Reddit, it's full of a, that's the cpit of the internet You really want that to be your training dataset There's a whole bunch of ethical dilemmas that I'm saying The, the thing I would sort of say is, and I learned this, this analogy from, , , Dean, , Pedro Vieira at Nova Business School in Portugal What he said is that, , AI is the wind.

You can't fight the wind We have to figure out how to work with it, ? And , yes, there'll be bias, it'll be this and that We have to make them better That's all gonna be all of our responsibilities But what I add to that is that we need to become the windmills, ? We need to be, to be able to harness it positively, ? To generate the energy we need, but we can resist it.

This is, this is, this has left the, , the horse has left the barn, or whatever analogy you wanna say this is, it's, it's just, it's just happening now And we just have to learn to accept it and then work towards changing it, work towards making it better And all of us should be feeling empowered about it, because guess what, with Microsoft Bing's, , chatbot, a free chat G P T service is available to every single user in the world It's incredible.

In fact, , in fact, , , open ai, if you want to access, , A G P T 4.0, you have to sign up for open AI's, , plus service for 20 bucks a month But binging offers you free access to, , Exactly A g PT 4.0 Exactly , , going back to this question about, , risks of ai, I'm really glad that the audience reminded us that let's not all be positive I think, , Dan Kari and I are just naturally positive people.

we look on these, yes, I'm An optimist, But, , we're all optimists But that said, I think we have to be very careful about, , the dark side of AI as well Where there are nights when I wake up, , I in a, in a, in a deep sweat, , , a society could get an highlighted, , it's one of those movies planted the Apes or something where, where suddenly, , , who knows what can happen, , in a very, very bad way for humanity Now, whether that's a, , , is gonna happen or not, who knows? But there are many steps we can take to, , to do things in the way , for instance, in the AI master's degree from UT Austin, a unit on ethics, , of AI is part of the required sequence of the program.

And, , in many of our courses that we teach, ethics is kind of part and parcel, , in the work that we do at edX We've also come up with, , seven principles for thoughtful use of ai , and if you google, , thoughtful use of ai, edX, you, you, you'll, you'll probably find, , the blog post We talk about how we want to be sure that whatever we develop with AI is gonna help humanity and not hurt humanity Second, any, we need to be protective of human data.

, can you imagine if someone accidentally uploads all the social security numbers into AI models? These models are complicated and intertwined and intense , I don't know if we can tease out all of that personal information we have to be very careful, , in the data we share we come up with seven principles, , and all the work that we do, we try to abide by those principles And the first principle is that as a scientist, as an engineer, what are, whatever we build, let's be sure that it is for the good of humanity.

And the second one is that whatever we do, let's be transparent about it Let's put it in the open source that people can review it and make sure it's not gonna harm I think as long as more and more developers abide by some of these, these principles, , I think, I think we can be a AI can be more and more a force for the good, rather than going down the dark side That's definitely a great way to end it And I wanna thank both of you for all of your insights.

You really made, , this information just practical and hopefully pushed everyone on the call to educate themselves, start using these tools that they can advance, stay relevant, and as we found in the study, make more money and create more opportunities , I really appreciate time for everyone listening and even those who don't join us today, you'll be getting the full research report you can go deeper into all of this great data to see, , what else can be very beneficial to you and your role and your organization And as was mentioned earlier, there's a ton of great edX relevant courses to, to this the edX, , catalog and academies of courses around AI include the edX course, prompt Engineering, which was mentioned earlier, and advanced chat G B T that course could really, really help elevate you when it comes to the learning these skills and Harvard X's Computer science for artificial intelligence course.

And to get more information about these courses and everything else, you can go to discover dot edX edx.org/ai and that'll be in the chat as well, you can just click over to learn more thank you much again , I really appreciate your time This was an unbelievable webinar, one of the best I've been on and moderated ever, and, , I've learned a lot I am sure the audience has.

And, , we look forward to continuing the conversation..

‍