I'm a believer in this technology. I believe that the cheaper it is, the more people will use it. So, even if we assume that models will get smaller for the same level of intelligence, the demand to actually have more compute is going to be enormous. And the reason why you want to have the compute close to you is not only in terms of performance. As a national security threat, I'm not saying this will happen, but it could happen that China or US or whoever decides to stop exporting compute to the rest of the world. If that happens today, Europe has zero AI at the scale of data the data center.
For over a decade, technology has promised to make life easier. Social media was meant to connect us, we were told smartphones would simplify our lives, and now artificial intelligence promises to lift even more daily burdens. Yet these advances also raise an important question: when do tools enhance human creativity, and when do they replace the very experiences that give life meaning? While AI holds immense promise, many people face challenges coping with current technological advancements that ‘redefine’ human communication, work, education, and democratic values. Today, we examine societal and cultural impacts of AI, identity, inclusion, governance, education, and community well-being. And are speaking with Nacho de Gregorio, author of TheWhiteBox, one of Medium's most-followed AI analysts, for a level-headed conversation about both the remarkable potential and the real limitations of AI.
NACHO DE GREGORIO
Thank you. I'm really glad to be here.
THE CREATIVE PROCESS
Well, we're glad to learn all you know, because it's such a fast-changing landscape. I can't believe, just a few years ago it wasn't like this. So you've spent years explaining AI to both technical and non-technical audiences. For you, what is the single biggest misconception about AI that has actually gotten worse maybe? What are people focusing on that doesn't matter nearly as much as they think, and what are they ignoring that you think really matters a great deal?
NACHO DE GREGORIO
Oh, wow. That's a really good question. I would say that the biggest misconception is this idea that AI is going to take all of our jobs. There's no evidence at all that this is going on, and I can get into the reasons why this won't happen, at least in the short term, due to cost. Meaning AI in several instances is way more expensive than the alternative in humans.
But if you analyze the technology from a more technical perspective, there are problems everywhere. It's not reliable enough. It has several other issues that, you know, it doesn't show any reason why we should suspect that this massive job displacement that we're seeing, the claims and so on. There's a famous quote by Charlie Munger that is, “Show me the incentives and I will show you the outcome.” If you look at the people that are saying these things, they are mostly people that have massive bets on this being true. Investors, founders of these AI companies that are trying to disrupt. Of course, they are always in this never-ending spree of raising capital, so they have to be really strong with the claims. But if you actually look at the numbers and see how the penetration of this technology on companies and in general in society, it's far more limited than we are told. And also if we look back in history and at previous disruptions, let's say electricity, PCs.
There was a really famous quote with personal computers by Robert Solow. It said, “PCs are everywhere, but in the statistics.” Meaning, these disruptions are—you can see the value in the technology, but they take time. PCs took more than two decades to really hit the spot in which it was an impact on society as a whole.
Electricity took around four decades. The steam engine took almost a century. So these things take time and there is always this idea that this time is different. Who knows? But right now, if we look back at history, there's no precedent at all. And yes, revenues are growing faster.
You're seeing some big winners in theory already. But again, I talk to companies for a living, and I really struggle to find examples of executives and companies that have adopted AI to a scale that really warrants the type of discourse that we hear by the so-called experts that claim that the world is ending.
THE CREATIVE PROCESS
So what do you say to people who are knowledge workers? They're coders, they're lawyers, they're people who—obviously journalists, that was already going down because of the evolution of the media landscape—and teachers. What do you say to those people who have lost their jobs or have had to reskill and adapt so quickly? Or maybe they haven't lost their jobs, but they are feeling the burden of their colleague losing their job, so now they have to run on this hamster wheel. So I'm not into the negativity, I just really want to know how we can prepare because even if it's not tomorrow and even if it's not next year, we are biological beings. Of course, we're not AI robots and we're not built to accelerate at those speeds. So how do we cope with it? What is your answer to them?
NACHO DE GREGORIO
For sure. To be clear, I'm not a Luddite. I'm a really big power user. I use the technology a lot actually, and I always incentivize people to adopt it. There's this famous saying that is, “AI won't take their job, but someone who uses AI will.”
Adopting technology is something that I would encourage everyone. It's a skill that has to be developed. If you look at the layoffs, most of the layoffs have nothing to do with productivity caused by AI. You clearly see there's something else going on, and that something else is just these people had to make a decision to either accept that they made a mistake or blame this magical thing. But yeah, it's true that for some cases, software developers, the job is changing. I don't want to sell this idea that things are going to stay the same way. It's being adapted to this new way of working. Again, software developer is a good example. You were saying lawyers, perfect example too.
You can do more with less in these places already, for sure. What I'm saying is that this idea that you can use AI in a business case to displace humans, it doesn't make sense. And the reason it doesn't make sense is because, as an automation tool, it makes far too many mistakes.
There's a thing—I don't want to get too technical—but there's a thing called the nines of reliability, meaning you need to offer a certain level of reliability to your customer. For example, with the cloud, with all these apps we use, Instagram and so on and so forth, we have very high requirements, around five nines.
This means 99.999% accuracy or uptime in the case of clouds. For agents, we're just on that first nine. This sounds great, like, “Oh, an agent is accurate nine out of ten times.” That seems good. At scale, it's terrible because the amount of mistakes that occur over a period of time is enormous. So there's a reliability problem there. And also the cost. You have NVIDIA themselves, who are at the center of all this boom, one of the more senior leaders acknowledged that they were spending more on AI than in humans. So there's a lot of this discourse saying AI is cheaper, therefore I can substitute humans. Not true. And even when we look at the technology itself, it's more of a copilot. It's more as an augmentation of human capabilities than something that you can meaningfully use to displace people. So I will say, however, that yes, I do believe that companies may downsize a little bit over a period of time. Diffusion will take much longer than people realize. But yeah, I believe margins will compress and over a certain period of time, companies will start to make these more. But what I'm completely rejecting is this idea that labor is going to end, that we're going to just go to the beach and have AIs work. That sounds amazing. I don't see any evidence at all of that.
THE CREATIVE PROCESS
And just the betting on it, I agree. It's not going to be overnight just because of the social unrest, but you raise questions about, of course, who pays for it. Out of all the billions that have been invested in it, my understanding is right now very few have actually recouped their investment...
