The problem with AI alignment is that humans aren't aligned
The problem with AI alignment is that humans aren't aligned
I'm sure there are some AI peeps here. Neural networks scale with size because the number of combinations of parameter values that work for a given task scales exponentially (or, even better, factorially if that's a word???) with the network size. How can such a network be properly aligned when even humans, the most advanced natural neural nets, are not aligned? What can we realistically hope for?
Here's what I mean by alignment:
- Ability to specify a loss function that humanity wants
- Some strict or statistical guarantees on the deviation from that loss function as well as potentially unaccounted side effects
You're viewing a single thread.
Some of the human-alignment projects look like "religions" and some look like "economies" and some look like "just talking to each other and trying to be halfway decent folks and not flipping out or some shit".
Heck, arguably the United Nations is a human-alignment project for x-risk mitigation.
Mmmm, agents training each other. Very Deepmind of you to mention that.
If you were doing your job and reading some web site, and you happened to notice that there were posts on that site containing child porn, wouldn't you hit the "report" button too?
Some of the human-alignment projects
And some look like "I flip shit bigger, align with me or I will flip your shit"
We'd like to bring all those and their existing versions together with the A-Better-World Consensus-Engine idea.
Tell me more about some of these other projects though please.