The (questionable) ethics of OpenAI.

AI engineering course at InterviewReady: https://interviewready.io/

I came across a video from Joseph Gordon-Levitt about data harvesting for AI, and it made me wonder...

Is it ethical to train models on publicly available (even purchaseable) data, and then cache it forever in a large langauge model?

After all, these models are similar to compression engines, which take data and try to condense them into approximate functions.

So the more you run these models, the more you use patterns which have been derived from someone else's work.

This isn't a problem with humans. We condense data into information all the time. We then generate art using pre-existing knowledge and styles.

But AI? It has infinite scale (practically). Is it ethical for corporations to pay just once for their training data?

Let me know your thoughts.
#LLMs #AI

Видео The (questionable) ethics of OpenAI. канала Gaurav Sen