Verified Human Sideloads for AI Safety: can a measured digital copy of

Raising $30,000active

I build "sideloads" - digital copies of real people with measured fidelity (two are running now). I want to test whether a copy of a specific trusted human can evaluate AI decisions at machine speed.

Project Details

I have my own sideload (mind model which is open sourced btw) and I will test it on evaluation of AI's decisions in some hypothetical situations. For example, if AI arrives on another planet and has small chance to destroy alien art work during landing - how my sideload will evaluate its decisions? Or some more existential human questions - like choices between love and safety. My goal is to find the most complex choices first where simple utility calculation will not work but the sideload can help to solve these problems.
I will do the work myself and the result will be publishable scientific article.

Theory of Impact

The results of the work can be immediately built-in into existing AI pipelines, especially as my sideload is open-sourced and can be used by other parties.

People

Alexey Turchin

Team Member

Grants Received– no grants recorded

Add grant

Funding Asks

grantmaking.ai Launch Round

Applied

Minimum

$30,000

Ideal

$40,000

What this ask is

How the money will be spent

New hardware - 5,000
Tokens etc - 5,000
Personal expenses - the rest

Discussion

No comments yet. Be the first to share your thoughts.