Lots of people talking about AI today. Yglesias, Krugman and Richardson all talk about the groundswell of voices within the AI companies themselves - even among the top leaders - for some form of regulatory oversight. The problem is that each company pursuing AI doesn't want to "lose" to their competitors. So each company rushes headlong towards faster and faster development and more and more powerful models.
When people like Jacob Coxon quit and go public with their concerns, it creates a moment when everyone takes a deep breath and looks around and ponders is maybe it's time to take a step back. The problem is that no single company wants to be the one to give up progress to a more heedless rival. Even these techbros are asking for a regulatory framework.
Back when the Food and Drug Administration and food inspections were created, lots of the bigger meat packers were supportive of it, because smaller operators were selling tainted meat and making everyone skeptical of processed meat. Same dynamic today. The main AI companies want an international regulatory framework to make sure AI doesn't get out of control.
As an aside, I don't think AI will "wipe out humanity." I heard a helpful distinction. AI is unlikely to produce an extinction event, but it will very likely create a catastrophic event. If a rogue AI took over air traffic control, hundreds of planes could crash, thousands would die and the global economy would take a Covid-level body blow. That would be really bad, but not the end of humanity. Just please keep fucking Grok away from the nuclear weapons.
This is a moment, though, where Trump's malevolent ignorance becomes potentially lethal on a large scale. His contempt for anything he didn't think of, his avarice to make a quick buck off AI stocks, his antipathy towards sharing power with other institutions means that Trump will actively thwart the steps that AI companies themselves want the government to take.
I remain amazed that he has a 37% approval rating. Seems high.