AI Agents Will Apply for Jobs And Make Money in 2025? | Microsoft AI CEO Reveals Future

you've been an inspiration a lot of the current AI revolution has to pay credit to you Mustafa suan Mustafa the CEO of Microsoft AI co-founder of Deep Mind One of the ogs of the AI World spent more than a decade at the Forefront of this industry before we even had gotten to feel it in the past couple of years now there's going to be a point in the future where these AI agents go out and make money by themselves just take a moment to think about how crazy it is that you can ask that question if anyone can go out there and just say hey spin me software that does ABCD what happens to the entire field of software as a service instead of just a computer being able to say things it will actually be able to do things it'll take actions it'll learn to use apis it'll buy things it'll write emails it'll make phone calls just as a entrepreneur would ladies and gentlemen today I'm

with Mustafa sulaman who you know to be honest you've been an inspiration ever since deep mind and I think uh you know a lot of the current AI revolution has to pay credit to you so thank you so much for doing everything you do and agreeing to do this very quick session I'm just going

to start with questions dive in the first question I had and I remember reading this one tweet about you a long time ago where somebody spoke about your version of the modern during test so you had spoken about the modern touring test where you said that there's going to be a point in the future where these AI agents they're sort of able to uh you know go out and make money by themselves right without too much human intervention and we've start to see some of that In Pockets right there's still some human intervention like we are a good example we use avatars on YouTube and Instagram and that's last month we did close to about 80 million views right at some point that's going to be an agentic process where there need not be humans involved my question to you is now that we're seeing some of this evidence what are your predictions now first of all just take a moment to think about how crazy it is that you can ask that question just really think people have spent 50 or 60 years being inspired by the touring test right and so I think many of your viewers will know what the touring test is but just for those who don't it is uh a long ago prediction by the amazing computer scientist and mathematician Alan churing who basically said that intelligence is when you can sort of convince somebody else that the AI system is a human and they can't really tell whether it's a human or whether it's an AI system because it speaks so fluently right that was the Turing test the imitation game and in the last year or so we've kind of just breezed past that moment right we've sort of passed the touring test and nothing seems to be uh you know nothing seems to have changed that's really profound um so like with that said I sort of you know kind of proposed a different version of that where um you know instead of just a computer being able to say things it will actually be able to do things it'll take actions right it'll learn to use apis it'll buy things it'll write emails it'll make phone calls um just as a entrepreneur would or an office worker would and um we're now at the moment where it hasn't quite approached that but sometime in the next few years you know you're right it's probably going to be the case that an AI could generate new videos on YouTube um try and promote them on social media to optimize their watch time um you know potentially try and monetize them in different ways maybe create merchandise for the show all kinds of things and that is a you know it is a pretty wild moment to be in it's amazing capabilities that I think are going to make us all much more sort of creative how does that change the economy I'm sure there's going to be a bunch of smart people that take advantage of this and a lot of other people who feel like they're being out competed Now by agents you had other humans to compete with in the past now you also have agents to compete with yeah I mean you know think about it like this I mean we've we all sort of adopt Technologies at different times um but net we all end up using televisions using cars having smartphones um and you know something that feels sort of profound and scary and uncertain and you know suddenly feels every day and mundane right it's kind of incredible that you know I can leave a voice note for my friend and dispatch that to another part of the universe it's incredible that I could have an idea and in two sentences visualize that idea in an image um and I just in general think sort of people underrate how profound the current moment is um and we get desensitized to um you know things that are already around us and are actually super significant all the time and so to be creative In This Moment is actually to be open-minded and to test and iterate and absorb and integrate and play with everything at our disposal and then you get to see the limitations of it right because if you don't engage with something you over idolize it as though it's like you know so much more than it actually is but when you really play with these models you see okay they're amazing and inspire iring in some ways but they're also super Limited in some ways and that gives you an intuition for how to use them when not to use them how to mitigate their weaknesses um and uh you know I think that's the sort of way to think about adoption I think one follow-up question

here is a lot of our audience is software engineers and a lot of the work we do in our company in the past we of course wrote code now we're increasingly seeing us use GitHub copy a bunch of other tools and starting to see that we're offloading some of our work there to the point where it feels like we're writing English right which is awesome in the first place but secondly what does that do to software because if anyone can go out there and just say hey spin me software that does ABCD and then just get the app or the tool and they're able to use it what happens to the entire field of software as a service yeah it's a great question and you know maybe if you look back um it's kind of easier to see that this has been the course of software development for many decades right we've been creating languages of increasing abstraction all the way from binary and Assembly Language up to you know now low code and no code languages for a long time and even when you know you write python you're really calling on a bunch of existing libraries there's many libraries that you're not writing from scratch every time and um so just as you strung together a whole series of different libraries now you're sort of giving a single instruction to compose an entire you know subcomponent of a program or maybe even the entire program itself so what is that mean it means that it reduces the barrier to entry to getting stuff done it makes it easier for anybody with fewer skills if you like um to write a piece of software um create an application design something see it produced in the real world what does that mean well that means that we have more experimentation of different ideas faster and so we're trying out as a species collectively all the different Paths of possible combinations of ideas that we could combine um and that in itself is going to increase the rate of discovery of new things new business models new product applications you know new scientific ideas which in turn is collectively going to drive the overall productivity of our civilization unlike anything before because after all invention has been the thing that has driven you know our species from the beginning of time to reduce human suffering to find shelter to make it easier to get food to reduce conflict and so that trajectory is about to go exponential another

question here is around you know cognitive effort right I was talking about this before the Pod went live where you I was looking at a really big number with a lot of commas and instead of me figuring out what that number was I just picked it up put it into an llm and said hey what number is this right I wanted the number in billions or trillions or whatever ever and it just gave me the answer you know that's something I would have figured out in 3 seconds but I didn't take the effort to take those 3 seconds I just dumped it in sometimes with YouTube videos I just pull out the entire video dump it into uh you know an llm and say please give me a summary of this so do you feel like at some point we're starting to get cognitively lazier and a lot of the tasks we used to do on a Microsoft Office or in Excel they kept the brain sharp right do you feel we're losing the bicycle for the mind you know throughout evolutionary history we as a species acquire Knowledge and Skills which enable us to adapt to whatever the environment requires of us and then we lose those Knowledge and Skills as you know civilization evolves like in the past it would have been important for us as foragers to know which berries were poisonous or you know which animals were likely to attack us or not right and we sort of drop that information over time it becomes less and less valuable to you know memorize you know certain pieces of information and more valuable to focus on creativity and judgment and so on so in know as I was saying sort of I think it reduces the barrier to entry to creativity right it make because I don't have to spend Years Learning to be a great programmer I can now try something out quickly then what the kind of evolutionary story is optimizing for is you know my ability to in invent a new idea A New Concept an abstract layer rather than my ability to be able to execute on it um so in a way it's sort of exercising a different part of our brain and yes atrophying a certain like for sure you know now that we all have smartphones we don't memorize telephone numbers like we might have done back in the day um or maybe we don't memorize our times table as much right or maybe we're a bit less good at Maps navigation because we all have a you know a map on our phone should do that affect education for youngsters now cuz we still do the time stable and we still do a lot of things that at this point none of us would do yeah you think education like especially kwell education has to change now that agent thing yeah definitely I mean I think it's already changing I mean if you compare it to sort of 20 30 years ago pre pre-digital it is a different educational environment to know that you now have access to information that your fingertips and obviously people worry that it means people are just going to sort of copy and not be you know um you know not sort of ingest or learn or memorize new information and that's going to remain important um but I think we just have to be kind of deliberate about it is a in a new phase every tool gives us an amplification of some skills and a kind of atrophying of other skills and we just have to sort of balance that in intention with one another interesting I want to

talk about Microsoft a little bit right I think you guys have a huge campus here in India in Hyderabad and you know something like 20,000 plus employees uh what is Microsoft's commitment to doing AI well in India like are there Engineers here who are contributing to copilot like what's the India play here yeah I mean actually even here in Mangalore I have team members who are contributing to co-pilot and contributing to our advertising stack contributing to our search relevance search quality um so you know this is some of the best engineers in the world India itself is one of the most popular uh markets for co-pilot it's growing faster than most other markets it's really quite impressive so it's a big part of what we're going to do

interesting I have one last question for you which is can you tell me the top three like from your you know backend data right what are the top three use cases of co-pilot today and then also paint a picture of where copilot goes three years from now and what that looks like what are the top three use cases you know in 2027 or 2028 yeah I mean top use cases are I think replacing your search engine yeah I mean many people are using it for everyday queries um you know what is the GDP of this or where do I find that or you know what did this person do you know generic queries that you might otherwise put to a search engine second is I think education you know it's definitely a help for college and school and third is I think increasingly I'm seeing people use it for companionship uh an emotional support helping to think through a tricky problem that you're working through because you know co-pilot today is not judgmental it doesn't put you down it asks you know it's always there to talk to you in a very sort of simple and calming way so I think those are the kind of big use cases but tomorrow or sort of in years to come um you know these models are going to sort of have near Perfect Memory right so they're going to be very uh useful as a kind of second brain for you like anything that occurs to you any idea that you have any open question historic documents that you have any tasks that you have it's really going to be a way for you to sort of augment your everyday thought process and brain and that's kind of what I mean by an AI companion that's always at your side seeing what you see hearing what you hear and living life alongside you do you think it's going to be a friend do you think I'd be able to give it a name because I do that a lot I spar with copilot a lot right which is hey I have this idea can you help me flesh it out and I know a lot of smart people that do that at some point you think I'd give it a name and it knows me very well and it becomes my friend I think it's going to become a friend yeah definitely you know really live life alongside you it's going to see what you see both in the physical world and on you know the digital world and the strange thing about it is that you're going to be able to sort of point to things like say you know take a look at this what do you think of that right and it'll just know what this or that is because it has a kind of presence living life with you it'll know what your style is what your tone is and how you like to talk do you think I'd be able to send it to work instead of me I think that you're going to apply for a job with it you know your personal AI is going to get to know you so well and fill in a lot of the kind of gaps that you have to allow you to be your best self you know it because it's infinitely malleable and adaptive you know and because I think we're designing it with such intention it should complement the areas where you deliberately want it to augment you to enable you to shine in the ways that you know you are sort of uniquely capable of doing and proud so I think of it as a kind of like jigsaw puzzle connecting with the areas that you choose to um you know sort of amplify in your life and yeah I can imagine you introducing it to your friends or to your parents or taking it to work with you or switching jobs with you Etc um I think it's going to be an important part of life so in a way nobody has to really suffer with loneliness like there's a huge loneliness epidemic right nobody has to suffer or at least has tools to help them out with that period and in the meantime I think it'll always have utility like this image was generated with co-pilot that's awesome so it'll always have cool image yeah it will always have utility I guess till that time Point very cool thank you so much

Mustafa this was like a super exciting session I had like these three four questions for you and I know we added a couple more questions because I wanted to explore some paths um I learned a lot and thank you so much for being here this has been great it's a lot of fun and thanks a lot cheers