The CEO of Anthropic published a major essay asking AI companies to voluntarily slow capability development; Microsoft signed on the next day, and Asian markets sold off hard this morning, which gives you a sense of how seriously the industry is taking it. This is just the tip of the iceberg. And, to cap it off, OpenAI decided to postpone their IPO until later (perhaps 2027).
There is so much to unpack here (which others are doing much more eloquently than I ever could), but I definitely have lots of thoughts on this. This “race to the top” cannot be sustained on so many levels. And, with the US-based AI companies pouring so much money and effort into this race, it’s truly getting a bit out of hand and blowing up that AI bubble to beyond bursting.
Open-source Chinese AI models are creating huge pressure, at least in one direction. Those models are more than capable of competing directly against the big AI companies and their LLMs here in the US. And there is the data center pushback across the nation. And there is this circle of money that seems to flow through all the major tech companies to support AI development.
The approach of AGI (Artificial General Intelligence) is only an excuse. It isn’t what is prompting this call for a slowdown. It might be part of it. And I agree governance is important, but I don’t see global governance coming quickly or easily. In the meantime, open-source models will keep improving and accelerating. Buckle up!