This is such an important question, initially from Cory Doctorow and then picked up by Alex Hanna. It is indisputable that it is a bubble which means that it will burst:
Doctorow thinks that the residue of the bubble popping will be minimal–large models will no longer be cost-effective to train, but small open-source models will remain, adept for smaller, better scoped tasks. If that’s all that the AI bubble leaves behind, then we’d be in a better place for society and science.
But I’m more pessimistic–and frankly upset–about what will be left behind once the AI bubble pops. Already, Google and Microsoft have sheepishly admitted that they are far from reaching their climate goals, due to the large investment in AI. Data center growth is putting immense stress on existing power grids, not to mention are turning literal Black bodies into grist for the mill for this insatiable machine. After the dust settles and NVIDIA has stopped churning out shovels (e.g. H100s) for the gold rush, what will be left behind? Will data centers go the way of shopping malls? Likely not–they’ll be repurposed for other massive computing projects. But what about those climate pledges? Will they be continued to be kicked down the road? To 2050? To 2075? Likely to some time which is too little, too late.
It’s not just the material infrastructure and the climate catastrophe, but the careers and industries which have been upset. Visual and conceptual artists have discussed how their work has all but dried up–to which OpenAI CTO Mira Murati wryly remarked “shouldn’t have been there in the first place”.
https://buttondown.email/maiht3k/archive/the-grimy-residue-of-the-ai-bubble/
I’m increasingly expecting the bubble will burst at precisely the point my book comes out 🤷♂️
Mark Carrigan 