AI is Deterministic Based Upon the Starting Data - AI Alignment Could Be Relatively Easy

11-28-2023 • Next Big Future - Brian Wang

An OpenAI employee has observed that Large Language Models starting with the same dataset converge to the same point. This would mean curating the data is the critical step in creating safe ASI (Artificial Super Intelligence). If we can front load the AI with the desired ethical and pro-humanity examples, then the resulting AI system should have a core of human aligned safety.

It could end up very easy to train the AI -LLM on the data of all the accumulated study of AI Alignment related knowledge and then the AI should learn and absorb the lessons.

interesting post from an OpenAI employee claiming that all large language models reach the same endpoint regardless of training strategy or clever tricks

this is of course what the bitter lesson teaches us but useful to get an up to date confirmation that it still holds true pic.twitter.com/kpBooKyqm4
— Andi ???????????????? (e/alb) (@Nexuist) November 23, 2023

1 Comments in Response to AI is Deterministic Based Upon the Starting Data - AI Alignment Could Be Relatively Easy

Comment by PureTrust

Entered on: 11/28/2023 12:07:07 PM

If AI really worked, and if the people behind AI were honest, they would simply develop AI for themselves, eliminate the need for 'employees' by using AI, live life better than ever without a bunch of complaining 'employees', and forget society. The reason they don't do this is that AI doesn't work, and they will use the high computing ability of AI to enslave people, because people work where AI doesn't.

Make a Comment (1) • Email Link • Send Letter to Editor