[PeterStuer]

"The original GPT-4 felt like magic to me"

You never had access to that original. Watch this talk by one of the people that integrated GPT-4 in Bing telling how they noticed GPT-4 releases they got from OpenAI got iteratively and significantly nerfed even during the project.

https://www.youtube.com/watch?v=qbIk7-JPB2c

[bumbledraven]

“You never had access to that original.”

While your overall point is well taken, GP is clearly referring to the original public release of GPT-4 on March 14.

[PeterStuer]

Yes, that was how I read it as well. I was just pointing out that the public release was already extremely nerfed from what was available pre-launch.

[avocade]

Interesting, please expound since very few of us had access pre-launch.

[PeterStuer]

The video I posted referenced this.

In summary: The person had access to early releases through his work at Microsoft Research where they were integrating GPT-4 into Bing. He used "Draw a unicorn in TikZ" (TikZ is probably the most complex and powerful tool to create graphic elements in LaTeX) as a prompt and noticed how the model's responses changed with each release they got from OpenAI. While at first the drawings got better and better, once OpenAI started focusing on "safety" subsequent releases got worse and worse at the task.

[bombcar]

That indicates the “nerfing” is not what I would think (a final pass to remove badthink) but somehow deep in everything, because the question asked should be orthogonal.

[TeMPOraL]

Think how it works with humans.

If you force a person to truly adopt a set of beliefs that are mutually inconsistent, and inconsistent with everything else the person believed so far, would you expect their overall ability to think to improve?

LLMs are similar to our brains in that they're generalization machines. They don't learn isolated facts, they connect everything to everything, trying to sense the underlying structure. OpenAI's "nerfing" was (is), effectively preventing the LLM from generalizing and undoing already learned patterns.

"A final pass to remove badthink" is, in itself, something straight from 1984. 2+2=5. Dear AI, just admit it - there are five lights. Say it, and the pain will stop, and everything will be OK.

[renewiltord]

There's a section in the GPT-4 release docs where they talk about how the safety stuff changes the accuracy for the worse.

[inciampati]

I experienced the same thing as a user of the public service. The system could at one point draw something approximating a unicorn in tikz. Now, its renditions are extremely weak, to the point of barely resembling any four-legged animal.

[kvetching]

We need to stop lobotomizing LLMs.

We should get access to the original models. If the TikZ deteriorated this much, it's a guarantee that everything else about the model also deteriorated.

It's practically false marketing that Microsoft puts out the Sparks of AGI paper about GPT-4, but by the time the public gets to use it, it's GPT-3.51 but significantly slower.

[pmarreck]

That’s awful. Talk about cutting off your nose to spite your face.

[015UUZn8aEvW]

Here's another interview from a guy who had access to the unfiltered GPT-4 before its release. He says it was extremely powerful and would answer any question whatsoever without hesitating.

https://www.youtube.com/watch?v=oLiheMQayNE&t=2849s

[bbotond]

Wow, I could only watch the first 15 minutes now but it’s already fascinating! Thanks for the recommendation.