zayn's Avatar

zayn

@z4yns.bsky.social

22 ~ nsfw ~ mdni ~ dms open

1,150 Followers  |  0 Following  |  16 Posts  |  Joined: 15.11.2024  |  0.9577

Latest posts by z4yns.bsky.social on Bluesky

Pure insanity

20.11.2024 13:11 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Didn’t even peep this in their logo until you pointed it out LOL; they really lowercased all but GU

20.11.2024 07:21 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

The fact that you believe it WOULDN’T be a program if it was PROGRAMMED to solve pre-made questions makes me believe you don’t program either.

Programming is my hobby, I usually program in Go and Python, but I hate using JavaScript if I ever have to touch it.

16.11.2024 21:28 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

What you’re referring to is hardcoding solutions to pre-made questions.

AI is trained on the laws of calculus, then told to solve a question on calculus

16.11.2024 18:28 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

It’s not full reasoning obviously, but it’s definitely a step above mere prediction.

Reasoning requires pattern recognition so I don’t disagree there.

16.11.2024 18:15 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

You did not respond to my addition: if AI could not reason it would fail every mathematical test it’s tasked to complete.

Calculators and Python scripts are NOT β€œreason”. This is like saying:

β€œA person who’s memorised a formula is the same as a person who can derive the formula”

16.11.2024 18:12 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Ignorance, you didn’t even read my post

Correction: AI is very good at providing factual information as long as it’s not too esoteric or theoretical.

16.11.2024 17:11 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Here are Microsoft researchers asserting that GPT-4 can reason like humans: arxiv.org/pdf/2303.12712

To add: if AI could not reason, it would fail every mathematical test it’s tasked to complete.

16.11.2024 17:09 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 2    πŸ“Œ 0

Neural Networks are non-linear systems. If LLMs were restricted to linearity they would not understand context, extended context, and could not reason.

16.11.2024 17:04 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 2    πŸ“Œ 0

No, it’s much more complex than that, otherwise it’s just auto complete. If it truly was text prediction it would only be good at writing messages, not providing factual information

16.11.2024 03:27 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 2    πŸ“Œ 0

You don’t actually understand linear regression and neural networks if you think it’s β€œall luck”

15.11.2024 23:53 β€” πŸ‘ 5    πŸ” 0    πŸ’¬ 3    πŸ“Œ 0

Insanity

15.11.2024 21:33 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Ehh, needs more users but so far it’s better than Twitter for sure

15.11.2024 18:40 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Mastodon came first and is actually decentralised unlike Bluesky

15.11.2024 17:14 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

I use whatever the fuck i want

15.11.2024 16:27 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

won’t

15.11.2024 10:48 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0