Introduction

OK, so today I am going to try and talk to you a little bit about the future of work. And I'm very specifically, I know I'm being a little bit more out there, degrees won't save you anymore. That's the title of the talk today.

This is a new one, so I'd love some feedback on what you think really lands and what doesn't.

The Future of Work

But it's really a story of how AI is reshaping work and why adaptability is now basically the most important skill that is out there. I'm going to start with a personal example, go through some practical frameworks and how we can think about this in the future, and then look at where that leaves us.

Practical Example

Now, I want to start with this. Many of you, not all of you, will have gotten this MindStone Connections invite. And I'm actually going to live trigger this right now so that you all see it as well. And you will see that it will make sense in a bit.

There we go. It's a very simple app. Oh, obviously, I need to actually put my Wi-Fi on because otherwise that doesn't work. OK.

So we've got the Mindstone New York February AI Meetup. What I can do here is I have the event attendees, which are all of you. I can map these columns, email and name. I import.

And let me see, I'm not going to select these. What I'm going to do, so I'm going to hit send invitations. All of you will, in the next minute, be getting an email that is about, hey, do you want to opt in to being matched to someone here at this event that is present?

You give it a little bit of information on what you're looking for, and it will match you up based on the people in this room that should really meet. And then make sure you find those people after this talk. So it's 400 of you that signed up today. So it is taking more than 10 seconds that I did the last time around, but there we go. All invitations should have been sent out there.

Building the App

Now, why did I start with this? There is a real story behind it.

1This app was built in 24 hours. And it started with an app called Bordi.

I don't know how many of you have heard of Bordi before? A few in the room. Bordi is basically this massive connector app that has been released a few weeks ago.

And you can talk to it. It asks you what you're looking for. It puts the data in a database. And then anywhere in the world, basically, it'll match you up when it finds a great match based on mutual connectivity.

Initial Excitement

I thought, that is great. We're running this massive AI community. Can I not do something like this for our community here? So I thought, amazing.

Started looking online, found a really great company that does these things. You can do connections within the community. And they did 85% or so of what I really wanted to get done.

Got really excited by it. Almost signed a $15,000 contract for our community, for all of you to get the connections that you really want, which would have been an annual subscription. I got really excited by it.

Building it Myself

Go to dinner with my wife and I talked to her about it because I was so excited about it. Halfway through me talking through what this was gonna be, I stopped and I'm like, shit, I think I can build this. And lo and behold, less than 24 hours later, I had built the entire thing using purely Replit and Cursor.

For those of you that know what those tools are, Replit basically prompt to app. And then Cursor, because Replit is great, but sometimes when you go a little bit more, when you really build a bigger app, it can get stuck. And I am still an engineer by background. I can use Cursor to then kind of get it out of that loop.

But I had built the entire platform in less than 24 hours. And now it was doing 100% of what I wanted, not just 85%.

Kudos to the founder of the other app that I had almost done. I told him, I'm really sorry. I know I told you yes on the phone, but here you go. I've got it working. I think this works better.

He jumped on a call with me, and he was like, thank you for running me through this. Shit, this is making me rethink what I'm building.

The Importance of Adaptability

But this is not just a one-off, right? By the way, take out your phone, try it out. I really want your feedback on like, is it actually giving you some of these matches?

The Half-life of Skills

Why this is important is that the half-life of skills has been going down for years and years. A few years ago, it was about five years, which means that the average skill you learn is valuable for about five years until it becomes not so.

Now, five years is already very close to the three years it takes you to do a university degree. It's just like you start learning at one, you do three or four years, and then you've got one year to try and really use that and somehow monetize it in one way or another.

I did my computer science degree. And so that five-year period, the two years you get out of it, if you're lucky and your degree is three and you get two out, that assumes that you're even learning skills that are valuable to begin with.

When I was going through my computer science degree, I was still learning Flash and ActionScript, which for like five years before that had already entirely been eradicated. No one was using that anymore.

And I do not think computer science degrees have gotten much better. You're not learning how to use Cursor today when you go through a computer science degree. And that is a real problem. It means you're already years behind before the time you go through.

Didn't get the email. Not yet? No? You didn't get email? Did other people get emails? Yes. OK, good.

So this is based on having signed up to MindStone, because that's the only entity list I have. But I'll have a look at it after. It's OK. So the other bit.

Personalized Tutoring

is that if there's one thing that we know, and Laura touched on it earlier, the Bloom 2 Sigma effect, the one study that we have on what is something you can do that has a dramatic effect on someone's learning outcome, it's personalized tutoring. It's something that is very directly related to each individual.

And so our ability to learn because we now have fully personalized experiences, also is dramatically empowered.

Now, we see this because I forgot to introduce myself properly. So obviously, at MindStone, we do AI training for non-technical people. For some of the biggest organizations in the world, we go in, we train their teams on how to use these tools.

in their job on a daily basis so they can actually get to the productivity ends, figure out which use cases they should use, how they can use AI assistance, stuff like that. Every day, I see real examples of going in and within four weeks, the entire team is 20% more productive.

And that is not me making things up. That is the teams reporting it back to us. If I reported some of the stuff we actually see, you wouldn't believe me.

We've had teams come back to us and say, we are now doing years worth of work in a few weeks. That is the reality in some use cases that I'm seeing every day.

And so that's going really fast.

AI in the Workplace

Now, the framework I want to leave you with is actually one of the AI personal trainer. Because previously, the way that we looked at learning is very much kind of one-to-many or even one where when it's a personal tutor, it's you learn first. you apply second, somehow. And then the tutoring is about learning, and then you somehow apply that in another way.

Generative AI in Gaming

At Krafton, we are revolutionizing the battle royale genre with generative AI. Introducing PUBG Ally, the world's first co-playable character designed to play like a human teammate.

Hey Ally, I'm looking for a level 3 vest and 5.56 ammo. Keep your eyes peeled. Found a level 3 vest and ammo nearby.

Paint it for you. Enemy spotted. I'm covering.

Is there a vehicle nearby? Found one. I will come to you.

I'm ready to get into the action now. Can you blink them? On it. Moving left. Wait.

I'm knocked out. Rolling to you now. Just hang on.

I couldn't have done it without you. I know. I carry for a living.

All right, let's see if it flips through and it doesn't. I have to actually flip in and out of this again.

AI Co-pilots in Work

OK, so if you didn't understand quite what was happening here, this was PUBG. So there was one player, and then you had an ally, which is basically a co-pilot. But the individual was basically saying, I need a vehicle.

The AI understands what that means, goes, scans the landscape, finds the vehicle, brings the vehicle back to the individual, and now I can start to use this. That paradigm, I know that this wasn't a game, but that is basically where we're headed from a work perspective, which is we all have these agents. We talked about it before.

Pablo mentioned that. And we're going to have these in our work every day. You're going to see 50,000 of them before the end of the year. This is the year of agents.

And each one of us in our respective jobs will have multiple to choose from to figure out which ones we want to use and others. But the flow is the same. you're orchestrating a set of agents to help you do a job in different ways, and they will execute on tasks in order for you to do a better job at what you're doing.

Corporate Adoption of AI

Now, this is happening already. This is an eye-opening one for me because we do these training sessions for corporates.

The biggest thing that I have is that corporates are moving really slow, or at least they're scared of the privacy domain, their security domain, and everything else, thinking they are protecting everyone. I did a session at Web Summit in Lisbon, 700 people in the room.

I asked, how many of you are using AI for work? 80% of the room had their hands up.

I then asked, how many of you are using your private accounts to go and do so? 60% of the room still had their hands up. 175% of people that are using AI for Work are using their private accounts to do so.

Why? Because they're not allowed to use these things by their corporate policies.

Corporate security thinks we're protecting the company by not allowing this. What actually happens is people use the exact same data, put it into a free ChatGPT account, and exactly what security and privacy and legal are trying to protect is happening because on those free accounts, all the data is being trained on. That is what is happening right now.

Challenges and Changes

And change is actually accelerating. Lots of people are talking about how we're hitting the limit on training these models.

Yes, OK, if you really want to be scientific from a computer science perspective, there is a particular part where there seems to be diminishing return, the actual training of the model. But when you look at what these models are capable of doing, when you look at how much of our tasks they can start to to execute on to a degree that we would be comfortable with, to a degree that we would be happy with. It's actually accelerating.

When you look at those metrics, so here this on the right hand side, you're looking at, was it this, it's a graduate level Google proof Q&A benchmark. Basically, it's how AI can solve different questions and it scores based on how many questions it gets right. This is what that graph looks like at the moment.

And so the capability of actually executing on human tasks of these models is getting dramatically better because we're finding other ways where we can improve these models, whether it's at inference or whether it's by building them into agents, a whole bunch of different ways that make them dramatically more performant.

Degrees vs Skills

So on that point, basically degrees at this point really don't matter anymore. I know that they're not going to disappear tomorrow, but in the next two, three years, whatever degree you have, that's never going to be the thing that will get you a job.

It is about, can you actually do the task that is right in front of you?

Conclusion

And the other bit is that luckily enough, AI in this case, when used well, really does amplify instead of replace. Like this whole co-pilot ally type paradigm gives all of us superpowers, so long as you understand how to use it, which is why, as Natan was mentioning earlier, it is so important that we actually help people understand how to use it.

Because if you don't know how to use it, then yes, you're in a situation where the job that you might be doing might indeed be automated. If you know how to use it, then it gives you those superpowers.

Now, this presentation was built entirely using AI. I literally fed all of the content I've been writing for the last two years into ChantiBT, asked ChantiBT what is the best thing that I can try and talk about, give me a presentation outline, did a bit of brainstorming with ChantiBT, fed that to Gamma. Gamma created all of the slides.

I basically did nothing other than, well, no, there is a lot of content that goes in and I do spend time here, but hopefully that was useful.

And on that note, You've got pizza and drinks.