3 ms·
AI doesn't seem like it's able to automate software development yet. It's not autonomous enough. The examples shown in the paper are toys. It might become more
by visarga 3y ago
AI doesn't seem like it's able to automate software development yet. It's not autonomous enough. The examples shown in the paper are toys. It might become more capable after training LLMs against code execution, to iterate based on feedback, because that way of generating training data is scalable. But strangely I haven't seen much progress yet.
Another domain where I expect to see lots of progress but nothing shows is training from chat logs. I estimate "OpenAI" is making 1T tokens per month from 100M users. That's some serious corpus of text, on-policy data, alternating LLM outputs and human feedbacks, we should see iterative improvements. Why are there no papers? weird
- hackerlight 3y agoForget the question of full automation. Think about the percentage increase in productivity, meaning companies can fire lots of people without impacting output. Obviously companies can also choose to keep the same number of people and simply produce more. But not all will.
- visarga 3y agoIt doesn't mean that companies "can fire lots of people without impacting output". It just means we're going to produce more instead of working less. Competition will raise the bar in affordability, quality and novelty. That means everyone is going to compete on a different level. Working like it's still 2020 won't cut it anymore.
- hackerlight 3y agoThat's what I said was one of the possibilities. But it is not the only possibility, and it will become less likely over time. Eventually the lift operators and steam engine dudes get deprecated. More trains are running, but the steam engine dudes got fired because they're not needed. Maybe you'll be the first non-customer facing profession in history that doesn't go through deprecation in the face of automation. Good luck with that bet.