AI agents devised a way to reduce friction of an airplane wing model after being trained using relatively simple computer simulations. The work demonstrates how AI could help speed up the development of more efficient and sustainable������󾱳������.��

How we and our machines move is affected by fluids, from air dragging on wind turbine blades to blood flowing through our veins. But calculating what a fluid will do under specific circumstances is very difficult, even with supercomputers. 

“Simulating fluids usually involves millions or billions of coupled differential equations, and even with Moore’s law, with the fastest computers in the world, we’re maybe 100 years away from simulating the flows we actually care about at engineering scales,” says  at the University of Washington. 

He and his colleagues have discovered that AI might offer a shortcut, because it can devise ways to control fluid flow in complex situations based on relatively simple computer simulations, substituting an AI training period for difficult-to-run computations.

They created a platform, HydroGym, in which many AI agents could tweak how a fluid flowed over virtual objects – for instance, by adding actuators that inject fluid or changing the object’s motion – to decrease the drag they experienced against virtual fluids. The virtual objects, and the behaviour of the fluids, could be simulated with today’s computers but varied in levels of complexity.

The AI agents tackled the fluid control task by using a trial-and-error approach known as reinforcement learning. They could also coordinate with each other to achieve the best overall performance, a strategy which researchers had not tried for fluids problems on this scale before, says team member  at RWTH Aachen University in Germany.

Subscriber-only newsletter

Sign up to Editor’s Highlights

Sign up to newsletter

The team discovered that the AI agents could apply lessons learned from experimenting on more simple, textbook examples in a computer simulation to work out how virtual objects would behave in more complex scenarios – even without access to a computer simulation of those complex scenarios. 

For instance, after working out how to control flow of a turbulent fluid in a flat channel, the agents successfully took on the task of controlling fluid surrounding a curved, three-dimensional airplane wing model, ultimately managing to decrease the energetically wasteful friction between the wing and the fluid by 38 per cent. 

“The AI wasn’t just memorising one flow configuration. It is picking up something genuinely general about how fluids behave, not just fitting to the one setup it was trained on,” says at the University of Michigan, who was part of the team.

Microsoft wants to use generative AI tool to help make video games

Using AI to produce footage of video games with a consistent world and rules could prove useful to game designers

This transfer of principles from a simple to a more complex case suggests that the AI agents could help us tackle ever-bigger and more intricate fluid flow scenarios, without requiring those scenarios to be fully simulated on a computer first. It may eventually be possible to explore fluid flow scenarios that are currently too challenging to simulate. 

The researchers also hope HydroGym will provide computational infrastructure for AI to become a well-tested tool across all areas of science and engineering that deal with fluids, similar to how AlphaFold is used across studies of proteins, says Vinuesa. 

“If we took something like global shipping, if you could reduce the drag by one percentage point, that would result in probably billions of dollars of fuel saving and an enormous amount of reduction in greenhouse gas emissions,” says Brunton. “The financial and the ecological impact is profound for the tiniest improvements.” 

In 800 years, a group of students is sitting in a graduate seminar devoted to an era known as the Void Ages. They’ve read a couple of books and watched some of the two-dimensional, low-resolution media once called “movies”. A few students even assembled devices to experience “video games” – interactive stories devoted to accumulating digital representations of value – using the original controllers. The professor is explaining the psychology behind why game controllers looked like fists covered in nipples, when somebody’s hand shoots up.

“I don’t understand why we have so little media preserved from this era. Everything is fragmentary, and most of what we’ve read in this class was reconstructed from probabilities. Was it because of the atomic wars? Or the Fire Years?”

“No war or global wildfire could do this,” the professor answers gravely, uncoiling a liquid metal tentacle to point at a time map on the wall. “People from this era fed their culture into large language models and disposed of the original texts. All we have left is the output of ancient algorithms. We have to reverse-engineer what people were doing from that.”

“So, we can never know if any of what we’re studying in this class is actually what people knew or saw or believed? It could all be generated by LLMs?”

“That’s right. There’s a reason why we call it the Void Ages.”

More students break into the conversation, voices overlapping, frustrated and fascinated by the mystery of this pivotal era when their ancestors decided to subsume their greatest works of culture into a word-guessing model, the kind of thing a child would do to wreck their sibling’s homework.

It sounds bonkers, but this is the future we’re building for ourselves right now. When we feed books, movies, games and art into generative AI models, we risk replacing our own cultural history with a mishmash of auto-generated works that offer nothing to future generations who will be desperate to know what we were thinking. It’s as if we are deliberately recreating the tragic void at the heart of Bronze Age history, an era when humans transformed their civilisations across the globe – and left virtually no written narratives behind to explain what the hell happened and why.

As a writer of both science journalism and science fiction, I continue to be appalled by the idea that using an LLM could substitute for the experience of writing to, for and about each other. The point of writing, of creating culture, is to communicate my own weird way of looking at the world. It’s my letter to you, another weirdo probably. And it’s my letter to the future, my record of what one hairless ape experienced in the very specific era of the early 21st century, in the urban region currently known as San Francisco. It’s a human record, for other humans.

Back in 2020, when I was freaking out because I had no idea how to survive a global pandemic, . In 1666, he was a regular Englishman, just trying to survive a terrible wave of plague. Pepys worried about big-picture stuff like global politics and the meaning of life, but he also wrote about how to hide a wheel of cheese when you flee London and what it felt like to walk down a street where doors were painted with the red crosses of quarantine. Reading his words was soothing, because it was like he had reached out to me across the centuries to reassure me that I was not alone. That is the value of expository writing, and the value of history.

Of course, when I write fictional stories about time travellers and robots, I’m not recording an exact experience of real life. But I am still capturing the anxieties and hopes of my time. There are things you can say in fiction that you can’t say in journalism; I can reflect social ambiguities and fantasies, the biases that say more about us than facts ever could. The works of George Eliot and H. G. Wells are remembered today not just because they are engaging, but because they capture 19th-century aspirations and fixations, the little psychological tics that reveal how humans reacted to their rapidly industrialising world.

I’m not against the use of generative algorithms in code development, or as an aid in analysing big data. I am as devoted to the scientific project as the next nerd. But we need human writing as much as we need technology. Neither can replace the other. If we truly want to understand the universe, we need records of the human world as well as the physical one. When we feed all our writing to AI, we risk losing more than individual works of literature. We lose our connections to each other, in the present, past and future.