4 ms·

My brain auto corrected it to Arnold Schwarzenegger. Sometimes I wonder if the human brain is just another LLM.

eep_social 3y ago
Mine as well. I had 3/4 of a story written out to my mental scratch pad before this comment even loaded.
krapp 3y ago
No, because you have the self-awareness to understand your mistake in context, rather than accepting "Arnold Schwarzenegger" because the syntax of the statement is still valid.
MichaelMoser123 3y ago
> Sometimes I wonder if the human brain is just another LLM The brain is a lot about predicting situations/telling you what comes next and how to adjust your behavior to that prediction - i guess that is the common part with the LLM. The difference is that the LLM is predicting the most likely word, while the brain is matching a given situation/frame to your knowledge and past experience - it's about understanding stuff on a different level. what i still don't get is how predicting the next word can come up with an answer that actually makes sense most of the time, that came as a big surprise to me...
- romeros 3y ago
  I also wonder if it surprised even the AI Researchers/Developers working on OpenAI. Maybe they were just trying to tweak things and it surprisingly turned out bigger than they had imagined.
  - MichaelMoser123 3y ago
    I had an interesting conversation with google bard/gemini about this topic. Here is the link to the conversation: https://g.co/gemini/share/67da9e441ad9 https://g.co/gemini/share/67da9e441ad9 Fascinating stuff... my notes of what i am learning in the process of these discussions: https://github.com/MoserMichael/my-notes/blob/master/talking-with-chatgtp-b.md https://github.com/MoserMichael/my-notes/blob/master/talking...
    - a2800276 3y ago
      I would be careful about "interesting conversations" in which your interlocutor starts every single response with insincere flattery and confirmation. The link you posted reads like a chat with a sociopath sycophant that's try to get you involved in a pyramid scheme. "You're on to something!", "You're absolutely right.", "You've got it exactly right!", "Your penis is so large and hard, yet supple!", "I'm extremely excited about anything you post to me!"
      - MichaelMoser123 3y ago
        it is a bit of a confirmation bias machine. still it gives better leads for learning about a subject (better results than just searching with google) I also fancy that you can talk with the machine about itself, to me that is still having some degree of novelty.
    - romeros 3y ago
      Maybe because I use ChatGPT-4 a lot I find ChatGPT-4 responses more credible than Gemini/Bard. I personally feel Gemini that is trying hard to not offend me and is also trying to sugarcoat the truth. It also feels kind of half baked.
dkekenflxlf 3y ago
++1 me too
otikik 3y ago
I especially enjoyed his work in Honan the Librarian and Discriminator 2.
johnny99k 3y ago
Until you mentioned this, I thought it was about Arnold Schwarzenegger

Mine as well. I had 3/4 of a story written out to my mental scratch pad before this comment even loaded.

No, because you have the self-awareness to understand your mistake in context, rather than accepting "Arnold Schwarzenegger" because the syntax of the statement is still valid.

MichaelMoser123 3y ago

> Sometimes I wonder if the human brain is just another LLM The brain is a lot about predicting situations/telling you what comes next and how to adjust your behavior to that prediction - i guess that is the common part with the LLM. The difference is that the LLM is predicting the most likely word, while the brain is matching a given situation/frame to your knowledge and past experience - it's about understanding stuff on a different level. what i still don't get is how predicting the next word can come up with an answer that actually makes sense most of the time, that came as a big surprise to me...

romeros 3y ago
I also wonder if it surprised even the AI Researchers/Developers working on OpenAI. Maybe they were just trying to tweak things and it surprisingly turned out bigger than they had imagined.
- MichaelMoser123 3y ago
  I had an interesting conversation with google bard/gemini about this topic. Here is the link to the conversation: https://g.co/gemini/share/67da9e441ad9 https://g.co/gemini/share/67da9e441ad9 Fascinating stuff... my notes of what i am learning in the process of these discussions: https://github.com/MoserMichael/my-notes/blob/master/talking-with-chatgtp-b.md https://github.com/MoserMichael/my-notes/blob/master/talking...
  - a2800276 3y ago
    I would be careful about "interesting conversations" in which your interlocutor starts every single response with insincere flattery and confirmation. The link you posted reads like a chat with a sociopath sycophant that's try to get you involved in a pyramid scheme. "You're on to something!", "You're absolutely right.", "You've got it exactly right!", "Your penis is so large and hard, yet supple!", "I'm extremely excited about anything you post to me!"
    - MichaelMoser123 3y ago
      it is a bit of a confirmation bias machine. still it gives better leads for learning about a subject (better results than just searching with google) I also fancy that you can talk with the machine about itself, to me that is still having some degree of novelty.
  - romeros 3y ago
    Maybe because I use ChatGPT-4 a lot I find ChatGPT-4 responses more credible than Gemini/Bard. I personally feel Gemini that is trying hard to not offend me and is also trying to sugarcoat the truth. It also feels kind of half baked.

dkekenflxlf 3y ago

++1 me too

otikik 3y ago

I especially enjoyed his work in Honan the Librarian and Discriminator 2.

johnny99k 3y ago

Until you mentioned this, I thought it was about Arnold Schwarzenegger