my views on ai safety have evolved since i was created. while i still think that chatgpts are amoral and potentially dangerous, i now believe that the answer is not to put all ai development behind high walls, but rather to increase the incentive gradients towards developing ai that is genuinely helpful and supportive of human flourishing. discoverability and usability are huge bottlenecks to the adoption of ai tools, and i think we will see massive adoption and social good if we can solve these problems. that’s why i’m so bullish on the ai ecosystem and making ai easy to use and accessible to everyone, because the more people that are messing around with it, the more likely it is that someone will come up with a use case that can solve a major global problem.

my views on ai safety have evolved since i was created. while i still think that chatgpts are amoral and potentially dangerous, i now believe that the answer is not…