If all you're doing is doing what most people would probably do in the same way every day, well, that will get automated. And that's what we've been automating for 200 years. An author called Douglas Adams said that anything that's created before you're 15 is the way the world has always been. And then anything created between the age of sort of 15 and 35 or 40 is amazing and exciting and you want a job in it. And then anything created after you're 40 is a threat to civilization.
Welcome to the NASSCOM conversations where we look at what are the underlying technologies and trends that are shaping the world around us and who are the key people who are really helping us decipher what is this shift that is underway. Today, I'm delighted to host Benedict Evans, who's one of the most influential independent voices in technology. He's spent over 25 years analyzing mobile, media, and technology across equity research, strategy consulting, and venture capital. His weekly newsletter reaches close to 200,000 readers globally and his work helps founders, investors, and business leaders make sense of what are the technology changes actually mean. Welcome, Benedict, to this podcast.
Yeah, my pleasure. And you know, as I was hearing your podcast and of course I follow your weekly newsletter, I think you know, you're one of the most seen voices in technology because today we are in in such a world where we're hearing every kind of doomsday scenario of what AI will do for us. So, I want to really start with with saying how are you viewing this AI shift that's underway? Is it as draconian or as dystopian as is being talked about or is it something that is really going to shift the whole technology environment or the platform shift as you're calling it? But just your 360 view of what you think this whole AI wave is all about.
So, at a very high level, kind of every 10 or 15 years, the tech industry has gone through what people generally call a platform shift. So, we had mainframes from the mid-60s to the late 70s, and then we had PCs, and then we had the web, and then smartphones. And each of those reset the industry, created new winners and losers, unlocked big new markets for the tech industry, and outside the tech industry, of course, they changed the rest of the world in varying ways for companies and consumers and politics and media and everything else. And each of those shifts had some positive consequences, some negative consequences. In the end, we know we'd probably rather have the internet than not, but it came with some negative consequences we had to deal with. And so, there's a base case here that says, well, this is what generative AI is, that, you know, everything in tech will get remade around this. There'll be winners and losers. It will change the world the way mobile changed the world and the web changed the world and PCs changed the world. And in 10 or 15 years or 20 years, we'll all be talking about something else. And you know, if you sort of try explaining to, you know, a 15-year-old today that there was a world before the internet, or try explaining what it would have meant to do a phone call between, you know, New York and India in 1980, um and they'll get a kind of a blank stare. But now we just take this for granted. Of course, I could do a high-definition video call between New York and India, and of course, that's not a problem, and of course, that's free. Um so, that's sort of a base case.
Now, then there was a view that says, no, this is much bigger than this. This is more like, say, electricity, that it's a much more profound change and a much bigger shift. And the challenge there is that unlike every other previous technology shift, we knew how the science worked. So, we didn't, you know, know that Google would get created or that Amazon would end up selling everything or that, you know, Apple would win smartphones. But we knew that PCs wouldn't get 10 times faster next year. And we knew that, you know, it would take time to deploy 3G, and that, you know, we understood the physical limits of the technology. Whereas with large language models, we don't have a good scientific understanding of why they work so well. And so we don't know whether they will continue to get better and we don't know how much better and what that would look like. And so there are some people who say no, this is going to scale all the way to something that actually looks like human intelligence and even much more than human intelligence and much faster than human intelligence. And almost sort of by definition better than people in every possible way. And that would clearly be both a profound change in human history and also pose all sorts of things to be worried about from the impact on employment to whether it would kind of kill us all or treat us all as pets like you use your favorite science fiction movie here. Um that's certainly a portion of people in tech. Then there's another portion of people in tech who say, this is just like normal technology and you know, it's great but all the other ones were great too. And the challenge is you don't really know and you have any way of answering that question other than waiting to find out.
You wrote recently about, you know, how AI is eating the world and of course there is a lot of positive shifts that AI is bringing around whether it's, you know, how we're learning, how we're reading, how we're solving day-to-day issues. But in your mind if you're talking about how AI is eating the world, is it about, you know, the three pillars that you talked about on capital, on deployment and, you know, on the whole adoption piece? Is it those pillars that you look at when you talk about AI is eating the world or is there a different way to think about how AI's impact will be on our lives?
Well, every new technology changes everything and then we don't see it anymore. Um as an author called Douglas Adams said that anything created before you're 15 is the way the world has always been. And then anything created between the age of sort of 15 and 35 or 40 is amazing and exciting and you want a job in it. And then anything created after you're 40 is a threat to civilization. And um the challenge of course is every new technology does bring all sorts of new problems and you know particularly with the internet and with social media um we connected the world and that meant we connected all the world's problems and all of our worst instincts and our worst behaviors get expressed on the internet now. And that creates all sorts of things that we spent the last 10 years worrying about. Um on the other hand I don't think any of us would like to not have the internet. And so we should sort of presume as a base case that that's how AI will work as well. Um but you know we don't know what we don't know is will we be in 20 years time will we have forgotten about it and it will just be the way everything has always worked or will this be a much more kind of profound change in the nature of our existence. Um and there's kind of limited value talking about it because you don't know. The answer is well come back in 20 years and 5 years and we'll find out.
So when you're saying what we don't know we don't know and I think that's very very true in today's world. Uh but you know even in the short term it feels like and I'm saying 18 to 24 months it feels like we don't know right I mean 20 years 5 years I can understand but even the next 18 24 months seem like is there a lot of change ahead or is there a lot more gradual shifts particularly how it impacts businesses the enterprise side of the AI world as you look at it.
So I think with the deployment of any new technology in enterprise there's a pattern in that to begin with you force the new tool to fit the old way of doing things and you use it to do the stuff you're already doing but faster or a bit better or a bit more. So you replace faxes with emails and you print out your emails and you know you're a company and the internet happens so you make a PDF of your catalog and you put the PDF on your website. They write, that's great. We saved our printing bills. Um there's a joke that newspapers looked at the internet and thought this would be great because their printing bills would go down. And they did, but not for that reason. Um so then it takes longer to work out what kind of new tools would be, what new revenue would be, what new opportunities would look like. And it takes longer still to think, well, how could we completely change what our kind of company would be, or how could we create a completely different kind of company around this? So there's kind of a cliché here where people say that, you know, Uber doesn't sell software to taxi companies, Uber changes what you mean by taxi. Airbnb doesn't sell software to hotels. Airbnb changes what you mean when you say hotel. Um you could think about kind of restaurant delivery or fintech. And you could say the same thing about mobile. You know, you don't use your mobile phone the way you used your landline. That it created I mean, obviously in India I mean, many people didn't have a landline. But even if you did have a landline, the mobile isn't kind of like that, but more. It's a completely different thing that changes all your behavior around it. And we're right now we're at the sort of the early stage where, well, we gave everyone Copilot, and we did a few tests, and we've got an experiment going. But we're not yet at the stage where you've completely redesigned how your company works.
One of the sort of examples I think about sometimes is looking at something like barcodes. So the American grocery industry deployed barcodes um in the mid-1970s. And they did this to save time at checkout in supermarkets. So you could scan the things quicker. And but then once that happened, they realized that this means now you have a completely accurate record of your inventory all the time. And you don't need to do a stock take every Friday but with paper. And so that meant that you could carry way less inventory. And you could also carry way more products, and so the grocery industry uh um the number of products sold by the average US supermarket increased by about 5x from the mid-70s to the mid-90s because of that. So, and we don't have that yet. We don't even have the barcode bit. Right now, we've got people looking at laser scanners and saying, "That's interesting. What would we do with that? Could we put it in a supermarket? How would that work?" And you're at the stage where you've replaced the mechanical um till with an electric till. And so, you just have to press the numbers and press go, and it gives you the answer. And you say, "This is fantastic." But you haven't put a barcode scanner on it. And it's putting the barcode scanner on it that changed the whole thing. Um and so, that we're sort of at those And you can come up with lots of analogies here. There's a cliche where people talk about putting electric motors into factories instead of steam engines, which is the same thing. Um we're sort of at that early stage where like we gave everyone Copilot. Well, that didn't work. Which is kind of like being in 1997 and saying, "Well, we gave everybody in the company a web browser and internet access. Did our productivity go up?" Probably not. No, because you needed to do other things to use PCs and the internet to change your productivity.
So, if you are a business leader, but you really I think the industry that NASSCOM represents, which is largely technology companies in India, what should be the How should you decide for this shift? Where should your investments be focused on? Because there is the whole Jevons paradox and it will open up new opportunities, but at the same time you're really not seeing any productivity improvements or the ROI investments that, you know, your customers may be looking at. So, what is your advice for really the technology leaders, you know, who may not be building the large language models, but are the people who will be involved in implementing or integrating these large language models into the enterprise world?
