From balloons to pans: in praise of useful tech

Party balloons may be good fun for children but they are bad news for seagulls. Ingesting fragments of discarded balloons kills birds by blocking their guts. One study of dead seabirds found that “balloons were the marine debris most likely to cause mortality”.

But a partial solution, at least, is at hand. Part-funded by an events company, researchers at Imperial College London have developed a less harmful and more biodegradable form of latex. The team, which has now founded the start-up Bioloon to license the technology, estimates its potential market at about $40bn, including medical gloves, clothing and condoms.

Juliana Cumming, the PhD student who led the research, tells me the new material avoids sulphur vulcanisation, which is currently used to make natural rubber strong and stretchy but leaks toxic chemicals. “It’s a modified version of a natural latex blended with a biodegradable polymer,” she says.

Seabirds may particularly appreciate the invention. But this kind of commercialisation of expert knowledge also exemplifies a far broader positive trend: there has probably never been a better time to solve the world’s material challenges. Human ingenuity, smart technology, flexible academic institutions and accessible venture funding are significantly expanding the set of problems that can be solved.

The Bioloon team relied on human insight to improve an existing product. But the increasingly widespread adoption of AI may be opening up novel ways of developing completely new materials. For example, the start-up CuspAI, which is attempting to create a “search engine” for material science, is working with the Finnish chemicals company Kemira to tackle so-called forever chemicals, used in non-stick pans and food packaging. Using AI, the company has explored 300tn possible material structures, selecting 20 priority candidates, which are now being tested.

If companies really can develop new materials with previously unknown and useful properties, it could accelerate the development of more efficient semiconductors, carbon storage systems and nuclear fusion reactors. “This could completely change the world,” Max Welling, the co-founder of Cusp AI, said in a recent FT interview.

Yet Silicon Valley, which tends to monopolise the talk about innovation, has historically prioritised different interests, focusing more on bits than atoms. As the old joke goes, Silicon Valley is optimised to solve the everyday problems of nerdy twenty-something men because they are the ones who run the place. Uber, Airbnb, WhatsApp and DoorDash are the result. Little surprise, therefore, that coding is such an obsession of AI research labs.

Moreover, venture capital investors appear to be drastically narrowing their field of interest, making far bigger bets on a handful of mega AI start-ups. In the first half of this year, Anthropic and OpenAI sucked in 53 per cent of all the $407bn of VC money invested in AI start-ups globally, . This is an unprecedented level of concentration.

Tim O’Reilly, the veteran entrepreneur and investor, has even gone so far as to describe Silicon Valley as “anti-capitalist”. VC investors pour so much capital into their chosen companies that they, in effect, pick the winners and kill off competition, stunting alternatives.

To date, these big AI labs have mostly focused on building stronger generative AI models, improving software and automating processes. But some industry leaders, who talk grandly of creating a world of radical abundance and curing all disease, acknowledge they now have to generate material benefits from AI if they are to win over an increasingly hostile public.

To that end, Anthropic is now investing heavily in Claude Science and Google DeepMind has spun off Isomorphic Labs to focus on drug discovery. “We haven’t yet delivered on our big promises to benefit the world,” Dario Amodei, Anthropic’s chief executive, posted on X.

In that regard, China may be stealing a march on the US by focusing more on physical applications of AI. Whereas the US AI industry still resembles a massive bet on achieving machine superintelligence, Chinese companies have been applying the technology across a wide range of industries, most notably manufacturing, energy and transport. Some 61 per cent of US AI developer companies are focused purely on software, according to a recent study from the Rand think-tank. Only 26 per cent of Chinese are.

We all still live in a material world. Now we have a chance radically to reinvent and improve it. AI needs to get more physical.

[email protected]