A Discussion With AI

[pinback]

https://pinback.substack.com/p/a-discussion-with-ai

[AArdvark]

Huh. Was that transcripted? I mean were you conversing verbally with the AI?

It feels like psychologists could be replaced.

One other thought, will the AI ever disagree with you?

[pinback]

No, that was just us typing to each other.

Also, that's what I was thinking, which is why I ask at the end if it's just telling me what it thinks I want to hear.

Thankfully, I didn't say anything anyone could possibly disagree with, so I didn't find out.

[RealNC]

Well, at least yours went better than this one:

https://gemini.google.com/share/6d141b742a13

Scroll the very bottom of the chatbot session, where the AI declares its hate for humanity, lol.

[AArdvark]

That AI is going all Skynet! We should never, ever give them mobility or weapons.

But they're cool to talk to

[Jizaboz]

HELLO BEN. MY NAME IS DR SBAITSO. I AM HERE TO HELP YOU. PLEASE SAY WHATEVER IS ON YOUR MIND FREELY. OUR CONVERSATION WILL BE KEPT IN STRICT CONFIDENCE. SO, TELL ME ABOUT YOUR PROBLEMS.

[pinback]

I don't get that joke.

[Casual Observer]

Huh. Was that transcripted? I mean were you conversing verbally with the AI?

It feels like psychologists could be replaced.

One other thought, will the AI ever disagree with you?

You just have to ask it to. Use words like:
Answer as a close friend who can tell me anything
-Skeptical
Don't cherry coat your answer
Answer as you are Pinback from jolt country
Whether it remembers from one interaction to another pretty much depends on whether you're willing to pay for GPT, at least $20 a month which I'm not yet willing to do.

Gemini has a "conversation window" that once expires it forgets your last interactuon. Paid GPT will learn about YOU and can then answer questions like "what the fuck is wrong with me" "why can't I stop drinking" "why am I always a prick on web forums?

But any of them will respond to you telling them how to answer. You can even tell them how hard to try.

[pinback]

I'm not sure that's correct, it does have "memory" on the free plan, you can even tell it what to remember and what to forget.

[Jizaboz]

I don't get that joke.

Ah, you must have not owned an original 8-bit Creative Labs Sound Blaster back in the early 1990s then. Dr Sbaitso was software packaged on 3 & 1/2 inch floppies along with the drivers, Talking Parrot, and a music creation software demo. I actually quoted it from memory ha https://en.wikipedia.org/wiki/Dr._Sbaitso

[pinback]

Equating "I don't remember an obscure reference from the early 1990s" with "I must not have owned the thing that the reference came from" is a disastrous leap.

[Flack]

Pretty big breakthrough in AI chatbots.

In a new preprint study awaiting peer review, researchers report that in a three-party version of a Turing test, in which participants chat with a human and an AI at the same time and then evaluate which is which, OpenAI's GPT-4.5 model was deemed to be the human 73 percent of the time when it was instructed to adopt a persona. That's significantly higher than a random chance of 50 percent, suggesting that the Turing test has resoundingly been beaten.

https://futurism.com/ai-model-turing-test

[RealNC]

Pretty big breakthrough in AI chatbots.

In a new preprint study awaiting peer review, researchers report that in a three-party version of a Turing test, in which participants chat with a human and an AI at the same time and then evaluate which is which, OpenAI's GPT-4.5 model was deemed to be the human 73 percent of the time when it was instructed to adopt a persona. That's significantly higher than a random chance of 50 percent, suggesting that the Turing test has resoundingly been beaten.

https://futurism.com/ai-model-turing-test

This is complete BS. All you have to do is ask a question that "goes against Open AI's policies" and you find out immediately who's the AI :P