4 ms·
I mean Gary Marcus and Elon Musk seem to agree here. AI has very suddenly become incredibly powerful and will be further weaponized. It should fall under some k
by wwarner 4y ago
I mean Gary Marcus and Elon Musk seem to agree here. AI has very suddenly become incredibly powerful and will be further weaponized. It should fall under some kind of regulatory framework for safety and security. By way of comparison, AI is far more dangerous than cryptography, which is regulated in a national security context. In addition, we should not underestimate the increasing power of AI in the near future. AI is already using AI to improve AI faster than people can improve it. Witness all the educators figuring out ways to identify fake writing, without realizing that they are just feeding data to the next revision of GPT.
- cwkoss 4y agoHow would you propose AI be regulated? I think any regulations would have to accurately encode our moral systems to accomplish safety and security without creating worse side effects, and since the law already does a pretty mediocre job of accurately encoding our moral systems, I'm skeptical that regulatory intervention can be fruitful. I fear that regulations' most likely outcome is increasing the chance of powerful AI being built as an amoral slave to capital rather than a societal good: it will be expensive to comply with regs, so it will kill mom+pop AI shops and only the megacorps will be allowed to play.
- wwarner 4y agoSomething along the lines of the norms and guidelines for CRISPR germ line editing