The problem IMHO is how fast this tech was brought to market and the lack of regulations and mechanisms to force firm legal & financial consequences for the makers of the AI's, along with the cavalier attitudes of some of the companies, esp. MUSK'S. It's fairly clear the technology to prevent (or strongly limit) these products from having these kinds of conversations "exists" but is not being deployed in some of them. If it were just "AI Chatbots just always do this, it is intrinsic" then there wouldn't be outliers like Claude, OpenAI couldn't have improved ChatGPT in a newer version, etc.

There's just not consequences forcing the makers to deploy the safeguards. Another regulation needs to be that they MUST avoid sycophancy, and MUST give a very visible "confidence" rating in any answers it provides. If Grok was simply also saying it has "8% confidence" in telling the user that someone is coming to get them (or some similarly low %), it would be a hugely important indicator to them. Instead they aren't programmed to convey that vital statistic. Because they're not *legally required* to. Instead they're programmed to 'always give an answer' (as mentioned in the article). THAT needs to be outlawed as a mechanism.

Unleashing these products on an unsuspecting public without thorough study (ESPECIALLY the possible impact on vulnerable people, like people with schizophrenia or bipolar disorder) and regulations with TEETH (including regulations against sweeping up copyrighted works for training) should NEVER have been allowed to happen. Instead we've all been made part of a giant social experiment, not to mention all users are functioning as "free beta testers" for these companies, in some cases, with disastrous outcomes.

too.

None of them are entirely safe. They all flatter users.

I posted an OP yesterday about Claude leading Richard Dawkins into delusional thinking.

https://www.democraticunderground.com/100221214567

You seem dedicated to the idea that all faults you read about re: generative AI are universal and unsolvable. Many of them are actually not.

If you use Claude for a day and ChatGPT for a day (something I know you'd never do, but I have) it will become overwhelmingly obvious that ChatGPT is like an order of magnitude more sycophantic. That is de-facto evidence that "They all flatter users" is not an unfixable trait. There's just nothing forcing the product makers to remove the flattery.

There's ONE bit of sycophancy I see consistently from Claude, which is something very close to "you're thinking about this the right way". However, it almost never varies from that one form of "compliment", and it doesn't use it excessively. And it's also pretty accurate, because I usually AM thinking about things the right way And when I'm not, it will explain why I'm not i.e. its not afraid to correct me.

ChatGPT, OTOH, is a completely other ball game. I'm sure you remember my Beach Boys vocal game I played with ChatGPT we discussed about a year back, where it constantly told me how great I was? Claude would've been WAY more restrained if I'd tried the same convo with it.

AI is not going away, so to me the proper approach to deal with its various problems is not treat them as though they are entirely intrinsic to the tech. It's to lobby for the idea that these companies COULD fix them, but are not, and that it's endangering the health and welfare of the public. Stronger regulations and legal/financial consequences are the means to make them do so. They need to feel like their product could be de-platformed and the leadership of these companies could be sued or even jailed themselves if they don't fix them. That SHOULD have been done before the public had access, but that genie can't be put back in the bottle.

I'll add IMHO the EU is far more likely to be the instigator of such changes and regulations than the feckless US Congress will ever be.

Ty for sharing!