Humans lie, and LLMs lie, but less often. Also, most people's version of utopia has more to do with doing less mundane stuff than has to do with knowing more stuff
Hi I'm June, software engineer turned independent AI researcher (for now). I've been writing and experimenting full-time for about 5 months, and finally ready to share some of what I've been working on. Questions I've been answering:
Which would out-compete between humans, machines and human-machine combinations?
What are the limits of self-improvement and meta-cognition when complemented with AI?
How does a machine know? Do you trust or verify a machine's work?
Can you encode reasoning at the harness layer?
What are the necessary and sufficient conditions to intelligence?
How do episodes turn to memories? Can we encode that process?
When coding gets automated, what is a copyrightable expression?
I believe that like security research, AI safety research should be red-teamed and documented. I mostly write on my blog, but currently actively reaching out to other researchers who may be interested in some of my work. You can read more at june.kim
Humans lie, and LLMs lie, but less often. Also, most people's version of utopia has more to do with doing less mundane stuff than has to do with knowing more stuff
Hi I'm June, software engineer turned independent AI researcher (for now). I've been writing and experimenting full-time for about 5 months, and finally ready to share some of what I've been working on. Questions I've been answering:
I believe that like security research, AI safety research should be red-teamed and documented. I mostly write on my blog, but currently actively reaching out to other researchers who may be interested in some of my work. You can read more at june.kim