4 ms·
Everyone letting ChatGPT do their jobs, and the people training ChatGPT are too lazy to do the work themselves, they actually use ChatGPT to train ChatGPT... wh
by damnesian 3y ago
Everyone letting ChatGPT do their jobs, and the people training ChatGPT are too lazy to do the work themselves, they actually use ChatGPT to train ChatGPT... what could possibly go wrong in this scenario?
- brucethemoose2 3y agoI wouldn't say OpenAI is "lazy," but this is still an excellent point about LLMs. The researchers training the ones we use now are in a massive hurry. They aren't meticulously curating datasets, checking usage rights, hiring a huge field of human curators, or thoroughly field testing them. They are madly working on a new improvement, publishing it in a paper, then moving onto the next thing. This has been a cultural issue in AI land for years. These experimental models were never meant for the general public, yet here we are.
- qayxc 3y agoRelated: https://arxiv.org/abs/2305.17493v2 https://arxiv.org/abs/2305.17493v2 Basically, if this paper is right, 2021 was the last year usable training data could be pulled from the internet. It could be all downhill from hereon out, as the amount of AI generated content will soon overtake actual human generated and -curated data on the internet. So the party could be over soon ¯\_(ツ)_/¯