The Future of AI Explained: AGI, Robots, Leadership Crisis & What Happens Next

0:00 We're living in a time of immense volatility.

0:06 So many wars are happening, more than any other time after World War II.

0:13 Satellite states are being formed between countries with similar goals.

0:19 Governments having an extremely difficult time in the West to regulate AI.

0:26 Governments grappling with very difficult deficits that they must pay back.

0:32 This, ladies and gentlemen, is what I like to call the perfect storm.

0:38 Just like any other storm, if you prepare for it correctly,

0:42 then you will come out the other side better than everybody else.

0:46 But if you don't, and if you ignore it,

0:49 then you will come out the other side battered and bruised.

0:53 But there is light at the end of the tunnel.

0:58 Or maybe this is the eye of the storm that we're in today.

1:04 You see, AI has existed for about 70 years.

1:09 It started in Dartmouth College as a research

1:12 project by these gentlemen here on the screen.

1:17 This picture was taken about 20 years ago.

1:21 Unfortunately, none of them are still alive today.

1:25 But they did leave behind a legacy.

1:29 A legacy where traditional AI was eventually built.

1:33 Even after many AI winters,

1:36 where for periods of even 8 years, nothing truly really happened.

1:42 I remember when I was running some

1:43 of the world's largest AI programs at Citigroup,

1:46 United Health Group, and others,

1:49 it was very difficult to do anything really huge with AI.

1:53 We had to hire hundreds of data scientists, engineers,

1:58 and everything was really for the sake of science,

2:01 never really for the sake of return on investment.

2:05 I remember even for creating an algorithm for writing NDAs,

2:11 we had to train it on 10,000 NDAs over a period of 6 months.

2:16 We had to use the power of over 150

2:19 people in order to manually tag all those things.

2:23 All because we had to build machine

2:26 learning models and algorithms completely from scratch.

2:30 Of course, that project never broke even.

2:33 It was never profitable.

2:35 And that's the reason why most of you and most

2:39 of the world hasn't really heard about AI until very recently.

2:45 We now live in the world of generative AI.

2:48 Most of you probably remember ChatGPT3 when it

2:52 came to the scene almost 3 years ago.

2:57 But the invention of vectorization and transformer actually existed

3:02 about 18 months before OpenAI came to the market.

3:06 And this is something not a lot of people know,

3:08 but actually Google was the pioneer in this technology.

3:12 And they had this technology in their hands

3:15 for good 18 months before OpenAI came to the market.

3:19 And they never did.

3:21 The reason was they knew that this was going to be the end of the search engine,

3:26 their very business they relied on so much.

3:33 Already, we're starting to see a technology called super narrow AI.

3:38 What is super narrow AI, you may ask.

3:41 These are AI technologies that can do a specific thing better than any human.

3:48 I'll give you a couple of examples.

3:50 There's about a 120 super narrow AIs that exist today.

3:55 I helped develop 12 of those.

3:57 One of them was an AI that can read CT scans, X-rays,

4:03 any type of imagery scans for hospitals better

4:06 than any radiologist and better than any doctor.

4:10 Another example would be self-driving cars.

4:14 If regulators allowed them to use all of their features,

4:20 the average miles before they reach an accident is about a million miles,

4:25 whereas a human drives 100,000 km and has an accident on average.

4:30 So already, we have super narrow AIs

4:34 that can do specific things better than any human.

4:38 And it's a collection of this that's going to help us build AGI.

4:43 Now, I know that this term AGI is a little bit loose.

4:48 So a couple of months ago,

4:50 I did meet with Jensen Huang and some of the leaders of the LLM models

4:54 in order to help come up with what is the right explanation of what AGI is,

5:00 so that the World Economic Forum and other organizations

5:04 can then disseminate the right terminology for AGI and ASI,

5:09 artificial super intelligence.

5:13 I'll briefly mention the difference to you.

5:16 AGI is the textbook worm that remembers everything it's

5:22 ever read and is able to join up that information

5:26 in a way that it's able to disseminate information and push

5:30 it together so that it can then come up with concepts,

5:33 but not new concepts, existing human concepts.

5:37 The definition of artificial super intelligence, ASI,

5:41 is an AI system that can come up with new methods,

5:45 new approaches to science, to physics,

5:48 to maths that humans had not invented before.

5:52 It's creating new things that humans have not invented.

5:55 That is the key to ASI.

5:58 ASI is most likely not coming for at least another 15 years or more.

6:03 Some scholars say it will never arrive.

6:06 But one thing I've learned in my lifetime is you never say never.

6:11 So, how long is it going to take for AGI to arrive?

6:18 Next year.

6:21 Many of you probably thought AGI will be coming

6:23 in the next 2 or 5 years or maybe even longer.

6:26 But I would argue that we already have forms of AGI.

6:30 I get access to a lot of these systems 6 to 12 months before they go to market,

6:34 and already I'm seeing very powerful AGIs which are available today.

6:40 I'll explain that more a little bit later.

6:44 You see, we keep focusing on the word AI.

6:48 It is the sexiest topic right now.

6:50 But we need to recognize that in the world of technology itself,

6:54 there is an awful lot else going on.

6:58 We have incredible examples of sensors such as IoT devices

7:03 being connected to LLM models in order to do incredible analytics.

7:09 I was working with Shell just a couple of months ago looking at their 50

7:13 million IoT devices and how they could

7:15 manage offshore assets remotely using AI and robotics.

7:21 When we look at the advancement of data

7:24 and data exchanges and how well that's evolving,

7:27 we're seeing multiples improvement every single year

7:32 of how quickly you can exchange data.

7:34 Satellite data exchanges being online already with Starlink

7:39 and many others coming on board soon.

7:42 When we look at memory chips, processors,

7:46 all of them are getting better at magnitudes every single quarter.

7:51 It's not even every year yet.

7:53 Everything is accelerating in the world of technology,

7:56 and all of this is coming together to create the perfect storm.

8:01 Again, I think you should look at this as an opportunity and not as a threat.

8:08 Many great things will happen as a result

8:11 of all of these technologies converging together.

8:15 Last but not least, humanoid technology.

8:19 There's over 100 companies on Earth today with over

8:23 100 million dollars in funding for building robots.

8:27 We got Boston Dynamics, we got Tesla, we got many others with over 80%

8:32 of the components and manufacturing happening in China, interestingly enough.

8:39 Over the next couple of years,

8:41 humanoid technology will combine with AGI in order

8:45 to allow AGI to put the theory that it learned in books and websites

8:50 into practice into the real physical world.

8:54 And this is exactly what's happening at the moment with Nvidia.

8:58 Nvidia's second most important project after making chips is digital twin.

9:06 They've managed to copy the entire Earth.

9:09 They can copy physics.

9:11 How does wind work?

9:13 How do objects fall?

9:14 How do things break?

9:15 Everything in our physical world has been

9:18 copied already into digital twin by Nvidia.

9:22 Many companies in manufacturing such as Mercedes and many others have given

9:27 Nvidia the data it needs in order to completely resimulate their factories,

9:32 simulate how things should be done.

9:36 The reason why this is so important is because

9:39 simulation is going to allow us to build AGI.

9:43 And in a few moments,

9:44 I'm going to prove to you how AGI is being built and why it's already

9:48 here and why different versions of it is going to be coming out next year.

9:54 As I said a couple of months ago, I spent two days with Jensen Huang one-to-one

10:00 and with the heads of all the other large LLM companies.

10:04 And a couple of key themes came out from that meeting.

10:09 It is that in the world we live in today,

10:11 the media loves to tell us that we're going to run out

10:15 of energy because of the data centers we need to power AI.

10:19 But at least for the next 12 to 18 months,

10:22 we have enough energy to power all the data centers which are coming online.

10:26 This is not going to stop AI.

10:30 Many companies in that meeting also presented to us very interesting use

10:35 cases of how AI is being used in order to revolutionize their businesses.

10:40 We saw examples from oil and gas,

10:43 from manufacturing, from sports, from all walks of life.

10:48 And this year, you're going to start seeing a lot more

10:51 case studies coming out of how AI is driving real return

10:55 on investment and truly transforming different industries.

11:00 Another thing that we saw is that this year is really the year of execution.

11:07 A lot of executives of very

11:09 large companies have been promising their shareholders

11:12 tremendous value with AI but haven't quite delivered on it just yet.

11:17 And there's a few reasons for that, which I'm also going to discuss today.

11:23 I see a lot of executives losing their jobs this year because they're not able

11:29 to deliver what shareholders and others are

11:32 expecting from them when it comes to AI.

11:35 So, watch out for that.

11:36 The key thing that came out from our meeting is really that we

11:41 need very strong leadership in this time

11:43 of highly uncertainty and high volatility.

11:47 It's very important for leaders to lean

11:50 in, to understand what this revolutionary technology is,

11:54 and to reimagine their organizations, and lead the company towards that.

11:59 And as we'll see today, many leaders are failing to do this.

12:04 So, if you're heading up your organization,

12:06 one of the key things I would recommend is to really learn about this topic.

12:12 Don't push it under the carpet anymore.

12:14 It's here to stay.

12:16 And the reason why I'm saying that is that there's

12:19 over $3 committed to AI over the next 5 years.

12:26 Ladies and gentlemen, that's 3 000 000 000 000.

12:34 That's a lot of zeros.

12:36 That's a few more zeros than in my bank account and most of us,

12:40 hopefully, in this room.

12:42 If we stack up $3 in $20 bills, it will reach a height greater than 16,000 km.

12:52 The Earth's diameter is 12,700 km.

12:56 It can wrap around the world once and then some.

13:01 Just imagine that.

13:02 And this is why I put this cool picture for you to visualize it.

13:08 The reason why I'm telling you this is not to impress you with the number.

13:12 It's to make you understand that when there

13:15 is so much money being put behind something, there's only two outcomes.

13:22 One, it's going to be as magical and as amazing as we think it is,

13:28 and therefore the world would be transformed, it will change.

13:32 Or two, it's going to happen anyway because

13:36 investors have to get a return on their investment.

13:40 So, for those of you who think that AI is a bubble,

13:43 which it is, it will deflate a little, but it's definitely not going away.

13:49 If you're sick of the word AI as well, guess what?

13:53 More AI is coming, and the topic is going

13:55 to get even bigger as we reach AGI next year.

14:00 So, it's really important that you get real comfortable

14:02 with this topic and you actually start learning how it

14:05 is and what it can do for your organizations

14:08 and companies and for the sport of racing because this truly,

14:12 truly is transforming industries.

14:16 So, let's talk a little bit about how this is

14:18 transforming from having AGI into generative AI into AGI.

14:27 You see, generative AI is built by basically brute forcing a lot of data

14:33 on top of a lot of compute power

14:36 and then allowing the algorithm to find patterns.

14:39 And that's how generative AI is essentially formed.

14:42 That's why we have so much hunger for data centers and so much hunger for data.

14:48 But you see, AGI cannot be built in this way.

14:52 AGI needs to have context.

14:54 AGI needs to have experience of the real world for it to be truly AGI.

15:01 This is why this year, Microsoft and many other companies,

15:05 they've started to push out agents.

15:08 Some of you perhaps have heard of Claude Bot and some

15:10 other similar agents which are very hot right now in the market.

15:15 Agents that can do things on the computer screens for you

15:18 just like it was a normal person working on the computer screen.

15:22 So, what's happening right now is we're

15:24 getting generative AI to experience the real world.

15:28 Firstly, in computers, inside of the computer screen, through agents,

15:34 and then very soon through applying it to robotics.

15:39 Already, a lot of robotics companies are downloading LLM models

15:43 into robots so that when they experience the real world,

15:47 they're able to do it better than without it.

15:52 The production of very realistic and capable humanoids today

15:56 is about 30,000 robots being produced around the world.

16:00 This will go up to over 100,000 by the end of this year.

16:04 And as AGI already technically exists, and we'll talk about that in a second,

16:10 and it's getting more and more powerful

16:11 with agents learning the real world through computer screens,

16:15 once we start downloading that information

16:17 and that capability into humanoid robots,

16:20 they'll be able to apply that into the physical world.

16:23 We'll be able to have robots to do

16:25 firefighting and dangerous jobs in the beginning.

16:28 And then over time, that will translate into more

16:31 normal jobs as they get cheaper to produce.

16:35 But it is the combination of agentic through computer screens and downloading

16:42 these models into humanoid robots that AGI will be built over the next year.

16:50 I would argue that AGI is already here.

16:53 I'm sure a lot of you when you have a health problem,

16:57 when you're writing something in particular,

16:59 you're using ChatGPT or Gemini or something like that.

17:05 I hear a lot of people laughing.

17:08 It is in my argument that AGI is already here from a normal text point of view.

17:14 I mean, whenever I've had a health problem,

17:16 I've gone to my AI model first before going to a doctor.

17:20 And then when I turn up to the doctor,

17:21 I'm much more informed than I was before when I just had Google.

17:26 But as agents are being rolled out aggressively this year

17:30 by Microsoft and all of the key AI companies,

17:33 agentic technology is going to be combined with generative

17:38 AI technology to build a second version of AGI.

17:43 Already, we have agents that can do almost anything on a screen.

17:47 Right now, as we're sitting here talking,

17:51 my laptop is open in my hotel room where I have Claude

17:54 Bot working for me doing a couple of marketing things right now.

17:59 Unassisted.

18:01 And it's able to do very complicated things.

18:05 Once robotics production increases over the next year and we're able

18:10 to take these first versions of AGI and download them into robotics,

18:14 then that's how we're going to get real intelligence by the end of next year.

18:19 So, I hope that provides some context and a bit of a timeline for you.

18:24 I appreciate it's probably faster than you probably expected,

18:28 but true AGI will definitely exist by the end of next year.

18:33 It's almost to the point now where this is no longer a prediction.

18:36 It's just basically a timeline of delivery.

18:42 When we compare humans and robots, of course, there are many ways to compare.

18:49 Of course, a robot can carry more things,

18:53 can carry more weight, doesn't get tired, has less room for error.

19:00 But one of the key things that I think we should understand

19:03 about the difference between humans and AI is that AI has perfect memory.

19:12 I can't remember something that was said to me last week,

19:15 not word for word, but a an AI system can.

19:20 If I'm going to go to a board meeting and I'm given a 200-page deck to read,

19:26 I need to read that a week before the meeting.

19:29 AI can read that in minutes.

19:31 So, processing speed is at least 10,000 times faster than a human.

19:38 What I'm trying to get to here is recognizing that humans

19:44 and AI have different capabilities is extremely important because then,

19:50 through understanding these different capabilities,

19:52 you're able to redesign how your organizations and how your companies operate.

19:58 We give the right things to AI, which are high speed and need high accuracy,

20:03 highly repetitive, and we give the things

20:06 that require human interaction to humans.

20:09 If we combine those two the right way,

20:11 we can really enhance the customer experience and the human experience overall.

20:16 It just takes a little bit of imagination and a little bit of a design approach.

20:25 One thing that is happening, and there is many examples of this now,

20:29 is there is a convergence of industries.

20:35 I see a lot of companies who used to do

20:38 something traditionally starting to play in different markets now.

20:42 Take for instance some technology firms like Google, PayPal,

20:48 technology firms operating in the tech space, but also in the financial space.

20:55 Take a look at banks.

20:57 Banks who are traditional banks now operating

20:59 in the tech space, in the SaaS space.

21:04 You can argue that Formula 1, traditionally a racing a sport organization,

21:10 is now starting to also develop technology and new experiences,

21:13 immersive experiences, which are different to how it was traditionally before.

21:18 They're starting to branch out into a new industry as well.

21:21 You see, with the power of AI now,

21:24 you can build things extremely quickly and much cheaper than it was before.

21:29 And that allows you to compete in new markets,

21:32 to go to places you just couldn't go before.

21:36 If you have a little bit of an imagination,

21:39 you can quickly see things that you couldn't do

21:42 before is now possible because of the power of AI.

21:50 Of course, with all this power is going to come great change.

21:56 My team and I did a lot of research around

21:58 the types of jobs that exist in the world today.

22:02 And there is roughly 800 categories of jobs.

22:07 This will drastically reduce over the next 4 years as we reach

22:12 AGI mass adoption to about 100 different types of jobs around the world.

22:19 And these will fall under six different categories.

22:26 I'm sure as you see the categories on the screen,

22:29 you're going to be able to imagine which one that you're going to be in.

22:33 But the key thing I'm trying to get to you is change is coming.

22:38 Change in markets, change in how companies operate,

22:42 companies that used to operate in only

22:44 one field starting to operate in different fields.

22:48 All of that convergence,

22:50 all of that capability will result in a simplification of what humans do.

22:56 Humans are going to end up doing things with humans, for humans.

23:00 We're no longer going to be spending time

23:01 copying and pasting things and doing things manually.

23:06 All of this will go away over the next couple of years.

23:10 What will be left is jobs around entertainment,

23:14 focus on sports, on human capabilities, on leadership, the things that we should

23:20 have been focusing about in the first place.

23:24 If you look back to Rockefeller and Henry Ford era,

23:30 that was really the time when humans were made to be like robots.

23:34 One person puts in a bolt, another person puts in a mirror,

23:38 and slowly through a conveyor belt, we build a car.

23:43 But that translated also into the corporate environment,

23:47 where we work inside cubicles and teams, which are very well defined.

23:52 A team can only do a certain thing,

23:54 another team can only do another thing, and it's all connected.

23:58 I can definitely see how that Henry Ford conveyor belt

24:02 has been translated into modern corporate culture that we have today.

24:07 But all of that is about to change.

24:10 And that's why there's going to be a huge simplification of what humans do,

24:14 where we're going to finally take the robot out

24:17 of the human and let humans be humans, finally.

24:22 One thing that I'm very excited about is hyper-personalization.

24:28 We've always had personalization.

24:31 When we receive an email with our name on it,

24:34 we always thought that was super cool.

24:35 How did they know our name?

24:38 But now we're able to go into new levels of personalization.

24:42 We're able to send an advert for every different client.

24:48 Before, it would take hundreds of thousands of dollars to make a commercial.

24:53 Today, with two or even $1,000,

24:57 you can make the same type of commercials you were making before for $200,000.

25:02 And instead of taking months, you can do that in days now.

25:06 I'm working with a lot of institutions, some of them banks,

25:10 where whenever they roll out a new product like a loan product,

25:15 they're able to provide an advert for every single client.

25:20 I'll give you an example.

25:22 There's a client that's been a client

25:24 of a one of these banks for about 20 years.

25:27 They know that client very well.

25:28 They know all of their data.

25:31 And they know that this client always wanted to have a Porsche 911.

25:35 So when a new loan product came out, guess what the bank did?

25:39 They made an advert of this person going to the bank branch,

25:42 receiving a loan, walking out,

25:45 going into the Porsche garage, and then driving off with a Porsche 911 in Italy,

25:50 in Tuscany, and guess what happened?

25:53 That person went and took the loan.

25:56 So with hyper-personalization, by using AI to generate adverts and images,

26:01 you can create new experiences that you never could before.

26:06 That increases the rate of conversion.

26:10 That increases the rate of engagement.

26:13 And we already see a couple

26:15 of different racing institutions doing this very well.

26:20 So one thing that I would encourage you is

26:22 if you really want to make your institution stand out,

26:25 try to use AI to create that hyper-personalization experience.

26:29 And you'll bring people that you never had before as a fan base,

26:34 and you'll keep them engaged for much longer.

26:38 One very sad and very important topic is

26:43 going back to the leadership that I mentioned.

26:46 80% almost 80% 79% of leaders in major corporations today that run the world

26:55 have said that they don't understand AI enough in order to govern it.

27:01 Just let that sink in for a second.

27:03 80% of the people that run companies which run

27:08 the world don't understand AI enough in order to govern it.

27:14 If you don't know how to govern something,

27:16 you don't know how to set the strategy.

27:19 If you don't know how to set the strategy,

27:21 you don't know how to set the instructions and the goals

27:23 for your teams to go and transform your company.

27:27 So this is why, and also because I sit in a quite a few boards,

27:31 that I can tell you a lot of executive leaders are

27:34 going to be fired this year because they simply don't have

27:38 the imagination or the grit to understand how transformative this technology

27:43 is and how to lead their companies into the new era.

27:50 One thing that I really want you to leave this room

27:53 with is with the understanding that you don't have forever to transform.

28:01 I would say most of your organizations

28:03 have about two to four years to transform.

28:07 And why is that?

28:09 It's really because the people that you serve today, they are changing.

28:16 They are also transforming.

28:18 They want different things tomorrow than they did today.

28:23 And that is only accelerating with time.

28:26 There are new sports coming up every single six to 10 months.

28:32 New sports that understand attention, marketing, entertainment,

28:38 and they're taking the attention away from your potential new

28:42 customers that would like to go and see horse racing.

28:47 You're competing now with more sports than you ever were before,

28:50 and this will not stop.

28:53 So what you do as organizations to transform the way you market,

28:57 transform the way you create that immersive experience has to accelerate.

29:03 And I believe if you don't do this over the next 2 years,

29:08 you'll start to see a massive decline in the amount

29:10 of interest in horse racing as other sports,

29:14 new sports, become interesting and market themselves better.

29:21 So how do you go about this?

29:23 What do you do in order to become an AI first organization?

29:28 The key, of course, has to be education.

29:32 You cannot do anything else reliably

29:35 without understanding the topic well enough.

29:39 At the end of these slides, I'm going to leave two different QR codes,

29:43 a free guide and the newsletter that you can go

29:46 and learn from, which gets sent out every other day.

29:50 Over 100,000 people read it today and they find it very useful.

29:55 The other thing is, of course,

29:57 once you have the education, try to learn the skills.

30:01 Try to get advice on how to use AI.

30:05 Because even though you may have read about it,

30:07 you may have watched a lot of YouTube videos,

30:09 you're not going to understand how to do it at an enterprise level.

30:14 So, that advice can really help you accelerate your progress.

30:21 One thing that I'm very passionate about is you don't just

30:24 go and use AI or automation to just transform old processes.

30:31 You cannot be lazy when you think about the new version of your organizations.

30:37 You must look at everything that you do

30:40 and rethink what needs to be done by a human, what needs to be done by an AI.

30:46 And when it comes to tooling,

30:49 please don't just go and buy whatever tool you see in the market.

30:53 You have to very consciously understand what is that tool

30:56 for, what problem is it going to solve in my organization,

31:00 how is going to make the experience better.

31:02 And you cannot do that unless you have education, unless you have advice,

31:07 and unless you understand what is it you're trying to change in the company.

31:14 And of course, after you've understood all of that and you're

31:17 going to transform your organizations and the way you do things,

31:20 you need to have the right frameworks,

31:22 the right methodologies to do it so you don't repeat

31:24 the same mistakes as all the other companies before you.

31:30 Last slide here, we're going to have a little bit of fun,

31:33 but also a little bit of a look into the future.

31:38 Over the next couple of years, the stage that we live in right now,

31:43 I would say is a period of chaos.

31:48 We will eventually, at some point, reach 20% unemployment globally.

31:54 As companies go bankrupt, the companies who haven't transformed.

31:59 As students stop going to university as much

32:03 because education becomes free through AI models.

32:09 Governments in the West don't have a lot of control on what's happening with AI.

32:14 And actually, they don't want to have a lot of control.

32:17 They want the private companies to progress as fast as possible.

32:23 And this will also create chaos in the marketplace.

32:28 Hopefully, this period of chaos and immense change will start to get better

32:36 and will start to die down as people adapt to the new way of working,

32:40 the new way of doing things.

32:42 Lots of solopreneur companies are going to start

32:45 competing with companies which have 50 or 100 people.

32:51 I believe at some point we're going to need a new economic system.

32:57 If you think about it, you go to work, you receive money,

33:01 you then spend that money on goods and services,

33:04 those companies spend money with other companies and the economy goes round.

33:10 But in the very near future, you could use AI to print a new shoe,

33:16 a new pair of sunglasses using 3D printing.

33:19 What is the point of going to a shop anymore if you can make whatever you want?

33:24 So then, that breaks the economy of buying goods and services

33:30 when you can just go and get it yourself through your AI.

33:34 So, I believe a lot of countries are going to struggle with this.

33:37 How does the economic system work when the systematic

33:42 way of receiving money and buying things changes?

33:46 Ultimately, to end the predictions of the future,

33:50 I think we're going to reach a new height of human

33:54 evolution where we're going to take the robot out of the human,

33:59 let the human be human for once,

34:02 and then, just as the calculator allowed us to come up with bigger calculations,

34:08 to create new forms of engineering,

34:11 we're going to be able to do new things we've never done before.

34:14 I truly believe in the next 15 to 20 years,

34:18 we will find a cure for many diseases.

34:21 We will find infinite sources of energy.

34:24 We will find ways of exploring other planets.

34:27 And we will come together as a civilization around these very important topics

34:33 rather than fighting about things which ultimately

34:37 don't really matter for the human race.

34:39 Thank you.

34:41 [applause] [applause]