*slow clapping*
I'm actually quite interested in machine learning and generative models, specially LLMs. But... frankly? I wish that I was the one saying everything that the author said, including his dry humour. And more importantly, I think that he is being spot on.
People are selling generative models like they were a magical answer for everything and a bit more. It is not. It is just a bloody tool dammit. Sometimes the best for a job, sometimes helpful, sometimes even harmful. And the output is not trustable, and this is a practical problem because it means that you need to cross-check every bloody iot of the output for potential errors.
I think that I'll join in and drop my own "angry" rant: I want to piledrive the next muppet who claims that the current models are intelligent.
inb4:
- "But in the fuchure..." - Vomiting certainty over future events.
- "Do you have proofs it is not intellijant?" - Inversion of the burden of the proof. Prove me that there's no dragon orbiting Pluto, or that your mum didn't get syphilis from sharing a cactus dildo with Hitler.
- "Ackshyually wut u're definishun of intellijans?" - If you're willing to waste my time with the "definitions game", I hope that you're fine wasting hours defining what a "dragon" is, while I "conveniently" distort the definition to prevent you from proving the above.
- "y u a sceptic? I dun unrurrstand" - shifting the focus from the topic to the person voicing it. Even then, let's bite: what did you expect, F.A.I.TH. (filthy assumptions instead of thinking)? Go join a temple dammit. And don't forget to do some silly chanting while burning an effigy.
- "Ackshyually than ppl r not intelljant" - you're probably an example of that. However that does not address your claim. Sucks to be you.
Based on real discussions. Misspelled for funzies.