If there was one clear message from NEXT24, it was this: AI is plumbing. And that’s exciting. In fact, it’s energising.

You might be sceptical, at this point, if you weren’t in the room. After all, plumbing is boring. Who gets excited about plumbing? Well, a century ago, those who had to trudge to the bottom of the garden to answer a call of nature on a cold night got very excited about plumbing indeed. And that’s why when I say that AI is plumbing, I’m not being rude in any way.

Plumbing is vital, because of what it enables. A safer, more sanitary life. Or take the digital plumbing we built at the end of the last century: personal computers and the internet were the plumping the digital revolution of this century needed to happen. The internet was plumbing, not a product, even though AOL and Compuserve had a good old go at trying to make it one. They failed, and we built a new society on top of the internet’s plumbing.

AI is our new plumbing

And AI is our new wave of plumbing — or, at least, that was one of the core messages of NEXT24. General-purpose AI chatbots, like ChatGPT, could well just be the AOL phase of AI; an attempt to productise what is ubiquitous plumbing. The conference kicked off with one of those insights that seem obvious in retrospect, but which most of us had missed before: we’ve trained our LLMs on digital content, so all it’s good for is digital content.

What if you took that same process and applied it to the sensor data coming out of the physical world? That’s the question at the very heart of Ivan Poupyrev’s talk: what if we take the techniques we’ve used to create the current wave of LLMs, and use them to create ones that understand the physical world? The examples he gave were simple: spotting if a package carrying critical medical supplies had been mishandled, for example. But the potential that lurks in such models, in a world increasingly filled with sensors, is incredible. People have been talking about smart cities for years, but is this the technology that will finally unleash it? How about autonomous vehicles?

New, distributed energy

Could AI also help us reverse centralisation? The ability to centralise and mass-produce has been a hallmark of the industrial age, but the digital age opens up the possibility of reversing that. In particular, Arash Aazami pointed out that the green energy transition is opening up the world of micro-power generation. As people put solar panels on their roofs, and turbines in their garden, we create an electricity grid made up of hundreds of thousands of generators, or even millions, not a few hundred. Germany alone has half a million balcony solar installs.

That’s a nightmare to control if you’re human. AI can handle that sort of large data set effortlessly. Once again, AI slips into place as the plumbing that allows an energy revolution.

When AI is all upside

In the global south, they’ve already spotted this. As Payal Arora explored in her talk, they’re charging into AI with a more positive mindset than much of the West. With less traditional infrastructure to be disrupted, they see more clearly the opportunity in the emergent technology, and the downsides weigh less heavily on their minds. With less to lose, it’s easier to reach for a brighter future.

We’ve tossed around the idea before that AI is better characterised as Assistive Intelligence, rather than an artificial form. It’s an idea that Microsoft has implicitly endorsed in naming its AI Copilot. Each of these examples shows the potential of this assistive technology in enabling significant transformation of the way we interact with the world. It’s the technology that allows digital to escape the silicon boundaries it’s been imprisoned in, and start to reshape the way we interact with our physical environments.

Brain-computer interfaces? Yup, AI could move those closer, too. (And hopefully without the icky idea of electrodes intruding into our skulls…). If our current wave of computers are fundamental calculating machines, we’re starting to use them to create thinking machines, even if those thought processes are very different from the ones we use.

The new computers

And that’s why work like that of Andy Kitchen is so interesting: he’s making silicon/organic fusion that could open up whole new types of processing, and hence new types of AI. Despite some companies racing to productise AI — and OpenAI seems particularly guilty of this — AI is at its most exciting when we think of it as a toolset. It is one we can apply to any number of intractable technology problems, and potentially solve them for the betterment of our lives.

This was, perhaps, the true new energy we found at NEXT24. It feels like technology is opening up again. The past decade has been the tale of the appification of everything: all the computing tasks we do, compressed, divided and turned into discrete apps on our phones.

But AI opens that up again. Just as the internet unleashed a wave of experimentation that eventually led to that ultimate product/plumbing fusion, the smartphone, so, too, is AI providing us with new tools, and we need to figure out how to use them. I mused on my own site that, right now, too many companies are trapped in the efficiency/cost reduction view of AI. But it holds more potential than that.

The art of AI

That’s what made the panel of creatives using AI, hosted by our event curator Monique van Dusseldorp, so compelling. Here was a panel of people who had been using AI since it was known as machine learning, and found actually applications for this new plumbing, from helping seniors reclaim their memories, to reducing doctors’ triage times. We need to imagine what AI could be, as much as we need to figure out how to make money from it.

Just after the conference, the New York Times ran a profile of Jony Ive, probably the most consequent designer of the digital age. His iPhone defined the form and behaviours of the most important consumer device in decades. In among his other projects, who is he working with? Which client has led him to recruit many of his former Apple colleagues to work with?

At a Michelin-starred restaurant, Spruce, a few miles from Jackson Square, Mr. Altman and Mr. Ive talked about how generative A.I. made it possible to create a new computing device because the technology could do more for users than traditional software since it could summarize and prioritize messages, identify and name objects like plants and eventually field complex requests like booking travel.

Yup, OpenAI. Working out what to do with our new plumbing is the exciting field right now. And it’s energising.