this post was submitted on 11 Feb 2024
16 points (80.8% liked)
Futurology
1757 readers
225 users here now
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Hallucination isn't a solvable quirk of GPTs, its their function. You can't get rid of it by throwing more money at the problem, you'd need another idea.
There are tools to manage major hallucinations. More are coming. Automated fact checking, pattern analysis, multiple layer analysis, etc.
Yes, there are functional mechanisms that power hallucinations. Especially in the probability models. But there are some powerful tools automate analysis of the outputs and rework for accuracy. Those are likely to improve to eventually reach a level of trust that is sufficient for many business use cases.