“Arms race” is the wrong mental model for AI. Here’s a better one.
So…let me get this straight. The “commons” here is the potential profit of AI. And the “tragedy” is that google isn’t hoarding it and open source folks that are doing all the innovation are content to not make massive profits from their work?
If thats what this means, then I hope whatever their plan is to gain power or control over new AI tech fails. I hope the folks working on open source AI can separate their work from Meta’s original work enough that Meta no longer has any claim to it.
The tragedy of the commons is an economic and ecological concept concerning situations where private parties will overuse a common resource because private incentives and public interests aren’t aligned. For example, overfishing or carbon emissions.
In this case, the problem as articulated in this article is that each party in the AI gold rush - Google, OpenAI, Baidu, etc - has an incentive to rush their AI development without adequate controls so they can get ahead of their competitors, potentially taking us into dangerous outcomes in which one of them produces AI that has far-reaching harmful consequences for humanity. I guess the ‘commons’ here is the future of human society, which AI developers have private incentives to take for granted.
The solution proposed is to adopt many of the classic solutions economists have devised for tragedies of the commons - points 1-8 in the article - and apply them to AI development in the ways set out in the article.
@theinspectorst The real tragedy is that the business sector will probably beat the free software movement to achieving “A” “I” capability.
Some researchers seem to believe the opposite is true. I think it was Facebook who said they have no moat. Maybe it was google.
We have no moat and neither does OpenAI is the leaked document you’re talking about
It’s a pretty interesting read. Time will tell if it’s right, but given the speed of advancements that can be stacked on top of each other that I’m seeing in the open source community, I think it could be right. If open source figured out scalable distributed training I think it’s Joever for AI companies.
