The EU, as reported by Wired, is getting increasingly enthusiastic-slash nervous about making their own AI, or at least their own large language foundational models. They aren’t unique on this: both China and the US have earlier and better-funded initiatives, and even smaller countries and blocks have made at least performative gestures in this direction.
What’s peculiar about this, although not unexpected, is how disconnected this strategic geopolitical competition front is from the other strategic geopolitical competition fronts. Everybody considers areas like energy, advanced hardware, and biotechnology as lead-or-die areas, yet the sort of AI foundational models they aim at isn’t particularly aligned with them. This doesn’t make these models useless either commercially or geostrategically, and what you learn doing any type of AI can often be translated to others (from direct computational techniques to things like chipset design).
But it is not really optimal either! There are ways in which large language and image analysis foundational models are useful for biotechnology or, say, materials engineering, but they aren’t as useful as AIs for biotechnology or materials engineering, and taking leadership in AI for an area of science or technology is one of the best available ways to gain advantages in the area. Things like ChatGPT have proven so far to be a net negative for much of science: we need fewer and better papers, and we’ve gotten exactly the opposite. This doesn’t mean AI is useless for science — we know it’s not — it just means that if you want a geopolitically relevant competitive advantage in biotechnology, you don’t need to get the next ChatGPT, you need something that’s the next ChatGPT of biotech AIs.
Resources are always finite and there’s always less time than you think there is. Whether you’re a would-be hyperpower or a would-be unicorn, focus is necessary to win, and saying “AI” is far from being a focused strategy. Your strategic priority is the knowledge or resource battle you have decided you have to win. AI, the right AI used in the right way, is just how you’re going to.