Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Hmm, tempting to make a thing that just loops:

description = gpt4("describe <yesterday's output>")

and chain that into

new_result = gpt4("<yesterday's output> was described by a human as <description> but it should have been visibly apparent as a unicorn. provide SVG output that <extra blah blah>")

and post that as today's output. I wonder what we'll get. I'll do it manually.



This is the answer. The power of the hundred billon plus models is that they can exhibit in context learning. By providing more examples in the context the output improves.


Now that is an interesting thought!




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: