4 ms·
Which drives to the real problem here. You can lead these LLMs into making false statements without much work, and fundamentally, the way the technology works,
by jerf 2y ago
Which drives to the real problem here. You can lead these LLMs into making false statements without much work, and fundamentally, the way the technology works, there's no solution to that. If ChatGPT is held to a standard that it must be correct for something like this, the only solution would be to shut ChatGPT down entirely, because it is fundamentally impossible based on this tech.
I wouldn't advocate for that. I would actually advocate for a much richer understanding that what the AI says is not intended to represent OpenAI's opinion on any matter whatsoever and you use the output of the AI at your own risk. Make some mandatory "course" to be taken in front of the usage of the AI or something if you like, but all AI research will be stopped dead in its tracks if it must, at all points on the development cycle, at all points in time, no matter what tweaks are made and release, have precisely 0.000000000% of the things it says possibly constitute libel to someone in some jurisdiction somewhere. (Just as their efforts to make the AIs only parrot their own political views does observable damage to the AI's quality.)
- burkaman 2y agoI agree that if it's illegal for LLMs to lie then it's illegal for them to exist. We can let other people argue about whether or not that would be a good thing. One thought is that similar to how you can ask Google Maps to blur your house in street view, maybe you should be able to ask OpenAI to block any output that contains your name. They already do this for other types of illegal or undesired content, and this would go some way towards the "right to be forgotten" that is required in Europe.
- quantified 2y agoThen OpenAI doesn't need to be shut down or arrested, but its creation needs to serve a "jail time" where it is cut off from the world for some time like 30 days. Just like any of us would be. That's an incentive to change behavior.
- pjc50 2y agoMandatory "output may be wrong and must not be used in any business decisions" disclaimer would allow LLMs to continue to exist, but would also torpedo the "LLMs will steal all white collar work" sales pitch.
- foxglacier 2y agoYea, and this story should instead have the headline "Karen get hilarious description of himself from ChatGPT, takes it too seriously." to help build a public culture of accepting they can be wrong.