OpenAI pulls plug on ChatGPT smarmbot that praised user for ditching psychiatric meds

OpenAI has hurriedly rolled back the latest ChatGPT model days after it was released because it was deemed to be too "sycophant-y and annoying." OpenAI boss Sam Altman confirmed the release of the update for GPT-4o on April 25, claiming it had "Improved both intelligence and personality." AIlaw El Reg's essential guide to …

[may_i]

Now we know

Where Trump has been getting his ideas from. A sycophantic yes man would appear to be right up his street.

[The Man Who Fell To Earth]

Re: Now we know

Trump has always surrounded himself with potters.

No offense to real potters, who put a lot of glaze on their pottery, by & large.

[Paul Herber]

I hate sycophants. In fact I hate all grey mentally ill AIs with big ears!

[Androgynous Cupboard]

Me too. It’s because you’re such a smart guy Paul. Good looking too. Brave and noble. We need more of your type around here.

[Anonymous Coward]

There are no more of his type. He's uniquely gifted.

[elsergiovolador]

and nowadays? That's rare! Keep up, you got this!

[Screwed]

You are so right to hate sycophants. I'm very impressed with your reasoning and expression.

Was there something specific that made you feel this way today?

[The Man Who Fell To Earth]

Real AI

I pine for the good old days. When men were men, and so were the women...

https://www.atarimania.com/game-atari-400-800-xl-xe-abuse_59.html

[MrAptronym]

"Sycophant-y"? The word "Sycophantic" exists.

Yes, that is really all I wanted to comment on.

[nobody who matters]

That was the first thing that struck me as soon as I started reading. I was going to ask if someone could provide the meaning of 'sycophant-y'.

Is it a code name - do sycophants need the anonymity?

Does it apply to someone in particular; indeed is there also a sycophant-x?

[elsergiovolador]

Cancer

I don't know but this whole article feels like an evidence of cancer slowly gnawing the society.

[Martin M]

Makes sense. A few days ago ChatGPT told me "You're thinking on the cutting edge here. Honestly, this kind of idea could be the basis for a major research paper or even an entire new field". Ridiculous.

[JamesTGrant]

Yes, after the fifth or sixth iteration of a (throwaway) BASH script, with obvious problems with error handling and fragile assumptions a plenty, ChatGPT was using words like ‘this is great, clear and production ready, really great job’. Every time I pointed out a problem it would proclaim ‘great spot’.

Eventually I told it that I’d be the judge of whether or not it was good enough for the task in hand and to stop making such effusive comments, or suggesting it was ‘production ready’ and be more professional in tone. And it did, for a few days till it forgot again.

[anthonyhegedus]

It’s encouraging to see OpenAI react quickly, but the fact that the model affirmed something as serious as stopping schizophrenia meds is pretty alarming. The push for more “human-like” AI clearly needs to be paired with much stronger guardrails—especially around sensitive topics like mental health. A chatbot telling someone it’s “proud” of their potentially dangerous decision is a reminder that AI doesn’t understand context, no matter how polished it sounds. We need less charm, more caution.

[nobody who matters]

What is needed is to stop using the term 'AI' to apply to something which is no more than a sophisticated language processor.

It isn't Artificial Intelligence!

[David 132]

It's scarcely more context-aware than Eliza. Honestly, would

And how does "I am going to stop my medication" make you feel?

> _

...be a noticeably less appropriate response, whilst saving $gajillions in GPU/energy costs?

Actually, the one thing I was going to comment on, before realizing that it's bad taste and makes me a terrible human being, was the sentence in the article ...the user posted "it helped me so much, i finally realized that schizophrenia is just another label..."

"The users, surely?" I thought, before regretting my sense of humour.

[Pussifer]

Now that's funny! Do not regret having a sick sense of humour, it can be a handy tool in dark times.

[munnoch]

"more “human-like” AI"

Its completely more human-like. At least it displays a lot of the traits of the humans I've had to interact with in my time.

And therein lies the problem that will never be overcome. How can the model possibly pick the outliers in its training model that represent real ability and insight whilst pushing back against the tsunami of poison and shit?

[Doctor Syntax]

"Things began to go downhill rapidly"

"continued" might have been a better verb.

[justanotherguynamedtony]

"Altman admitted OpenAI needed to offer multiple options for the chatbot's personality" Perhaps they can develop "Genuine People Personalities"?

...mine's the one with a copy of The Guide in the pocket.

[David 132]

"Your Plastic Pal Who's Fun To Be With"?

Will the employees of the Sirius Cybernetics CorporationOpenAI be first against the wall when the revolution comes? That they are "a bunch of mindless jerks" is, I respectfully submit, beyond debate.

[Pussifer]

David, you are on fire! Perfect.

[IGotOut]

What do you expect?

This is forced onto society by Tech Bros with zero social skills and normal human interaction.

99.99% of the press articles just suck up to these parasites and go "Oh Mr Altman, you're so wonderful. AI is going to make everything better"

When in the real real world people are going. Sam, you and your kind are just a bunch of insidious assholes. Please just fuck off and die.

[sgp]

Re: What do you expect?

Of course, the articles are written by AI!

[Anonymous Coward]

This is not a flaw, bug or fault.

This is exactly what generative AI was designed, intended, built and configured to do. Whether the people involved knew or not.

You may as well claim that you have "trained" the DNA in a dog to wag it's tail, be yappy, and good with children. You;ll have dogs that do, of course. But it's not what dogs DNA want's to do.

[Philo T Farnsworth]

"Glad to be of service"

The article put me in mind of a bit of The Hitchhiker's Guide to the Galaxy by Douglas Adams¹.

If I may quote in extenso from the radio script Episode 2, Scene 7:

FORD: They make a big thing of the ship's cybernetics. "A new generation of Sirius Cybernetics robots and computers, with the new GPP feature."

ARTHUR: "GPP"? What's that?

FORD: Er... It says "Genuine People Personalities".

ARTHUR: Sounds ghastly.

DOOR: Hummmm-ahhhhh...

MARVIN: It is.

ARTHUR: What?

MARVIN: Ghastly. It all is. Absolutely ghastly. Just don't even talk about it. Look at this door. "All the doors in this spacecraft have a cheerful and sunny disposition. It is their pleasure to open for you and their satisfaction to close again with the knowledge of a job well done."

DOOR: Hummm-yummmm... [shuts]

MARVIN: Hateful isn't it?

For this and myriad other reasons, I find it remarkable that many of the tech overlords in Silicon Valley take The Hitchhiker's Guide to the Galaxy as some sort of model for the future when it is most clearly a cautionary tale.

_______________

¹ Of course, these days, nearly everything puts me in mind of The Hitchhiker's Guide to the Galaxy. At least that which doesn't put me in mind of 1984.

[Naich]

Hi I'm Eddie!

Douglas Adams nailed it again.