If this is the way to superintelligence, it remains a bizarre one. “This is back to a million monkeys typing for a million years generating the works of Shakespeare,” Emily Bender told me. But OpenAI’s technology effectively crunches those years down to seconds. A company blog boasts that an o1 model scored better than most humans on a recent coding test that allowed participants to submit 50 possible solutions to each problem—but only when o1 was allowed 10,000 submissions instead. No human could come up with that many possibilities in a reasonable length of time, which is exactly the point. To OpenAI, unlimited time and resources are an advantage that its hardware-grounded models have over biology. Not even two weeks after the launch of the o1 preview, the start-up presented plans to build data centers that would each require the power generated by approximately five large nuclear reactors, enough for almost 3 million homes.
TDD, Test Driven Development. A human writes requirements, with help of the AI he/she derrives tests from the requirements. AI writes code until the tests don’t fail.
Yeah, go ahead try that and see how it works out for you. Test driven development is one thing, having an AI try to write the correct code for you by blindly writing code is idiotic.
Why is it idiotic? Your tests will let you know if it is correct. Suppose I have 100 interface functions to implement, I let the AI write the boilerplate and implementations and get a 90% pass rate after a revision loops where errors are fed back into the LLM to fix. Then I spend a small amount of time sorting out the last 10%. This is a viable workflow today.
AI training takes forever. I dont think you realize how long an AI training actually takes. It’s not a 5 minute exercise.
Sure but the model is already trained. I’m not talking about using any sort of specialized model.