4 ms·

The value it haves for you may exceed or not its price. But "you" is an important part of that equation.

by gmuslera 3y ago

gumballindie 3y ago
What value does a procedural text generator provide?
- TeMPOraL 3y ago
  Depends on where it sits on a spectrum between, say, a dumb Markov chain, and a human like you and me. GPT-4 is definitely closer to the latter than to the former.
  - throw_pm23 3y ago
    Yes, but what do you use it for?
  - gumballindie 3y ago
    > GPT-4 is definitely closer to the latter than to the former. That’s a seriously dubious statement.
    - speedgoose 3y ago
      If you think about how people use and perceive the technology, it’s not that dubious.
    - MattRix 3y ago
      Have you used it? A GPT4 chatbot could convince 90% of people it was a human, right now. It could pass the majority of knowledge-based tests in all kinds of fields. 100 million people find it useful enough to use weekly.
      - martindbp 3y ago
        No level of capability is enough to convince some people it's not just a stochastic parrot. You're completely right on the Turing test. At this point it's just a matter of how long it could fool us. You have to remember it's not specifically trained to fool us either, it's trained through RLHF to be verbose and "helpful". If you train it right and add in some RAG (vector databases) I think you could fool most people for a pretty long time, some gullible people forever.
      - gumballindie 3y ago
        It could convince 90% of the people that are ready and willing to think it was human. Even when it does, it's only a piece of software designed specifically to emulate, therefore it's fake. Regardless of how much people want to claim it is intelligent, it's not. Sure it has load of data at its dispoal to generate well mimicked output, but to some of us it's so obvious and buggy that it's a waste of time.
        ben_w 3y ago
        If it is convincing most people, then surely by definition it's closer to us than to a Markov chain.
- HKH2 3y ago
  I don't work in tech and I use LLaMa. It is great for writing overviews of a subject which I use to make sure I have most of my bases covered (and helping me see which of my ideas are more original), getting more examples, along with other creative tasks. Getting it to compare two subjects to see what they have in common is usually quite helpful.
  - gumballindie 3y ago
    The issue is i tried using these types of tools for a while and each response is inaccurate. It becomes annoying to constantly ask it to correct it or having to doubt the output. Works well for what is statistically likely to be abundant in information but you know that’s usually of little value.
    - HKH2 3y ago
      I use them for brainstorming. I get such great ideas, I don't mind if they miss now and then. Information can be abundant yet hidden in obtuse or misleading writing. LLMs produce writing that is quite accessible.
- MattRix 3y ago
  I’m a little perplexed at why you feel the need to refer to it in reductionist terms like this? It seems to provide a lot of value considering that it has 100 million weekly active users. I regularly find myself using it a dozen times a day. I feel like the only reason someone wouldn’t find it useful is if they weren’t ever doing anything new.
  - gumballindie 3y ago
    Loads of useless products have hundreds of millions of users. Threads, Twitter, Facebook, and so on. Just spamming forums to promote this product wont increase its utility.
    - MattRix 3y ago
      None of the things you mentioned are useless. Perhaps you‘re just being needlessly critical? Or is this just a form of trolling?
    - Turing_Machine 3y ago
      Why do you get make the decisions about what other people find useful?
- ben_w 3y ago
  It can be an infinitely patient teacher for physics and German, that never gets frustrated by me getting the conjugations wrong or having whacky ideas like getting lift from plasma held in a magnetic field. It can write boring code, like the OpenStreetMap query to find all supermarkets within 50x50km of Berlin. It can do code and literature reviews of my solo projects and stories. It can take a list of ingredients and suggest things I can cook with them, and unlike "real" recipe websites it can do this without telling me how their grandmother used to comfort them with this on a beautiful autumn weekend, when… [1000 words and four ad breaks later] … and then when the website actually gets to the recipe it turns out to be unsuitable anyway. It can write form letters It can help role-play difficult real-life scenarios. - I don't expect perfection in any of these things. If it was perfect, us software developers would already be redundant, as would all the lawyers, the accountants, and anyone else whose job can be reduced to images and words.

The value it haves for you may exceed or not its price. But "you" is an important part of that equation.

gumballindie 3y ago

What value does a procedural text generator provide?

TeMPOraL 3y ago
Depends on where it sits on a spectrum between, say, a dumb Markov chain, and a human like you and me. GPT-4 is definitely closer to the latter than to the former.
- throw_pm23 3y ago
  Yes, but what do you use it for?
- gumballindie 3y ago
  > GPT-4 is definitely closer to the latter than to the former. That’s a seriously dubious statement.
  - speedgoose 3y ago
    If you think about how people use and perceive the technology, it’s not that dubious.
  - MattRix 3y ago
    Have you used it? A GPT4 chatbot could convince 90% of people it was a human, right now. It could pass the majority of knowledge-based tests in all kinds of fields. 100 million people find it useful enough to use weekly.
    - martindbp 3y ago
      No level of capability is enough to convince some people it's not just a stochastic parrot. You're completely right on the Turing test. At this point it's just a matter of how long it could fool us. You have to remember it's not specifically trained to fool us either, it's trained through RLHF to be verbose and "helpful". If you train it right and add in some RAG (vector databases) I think you could fool most people for a pretty long time, some gullible people forever.
    - gumballindie 3y ago
      It could convince 90% of the people that are ready and willing to think it was human. Even when it does, it's only a piece of software designed specifically to emulate, therefore it's fake. Regardless of how much people want to claim it is intelligent, it's not. Sure it has load of data at its dispoal to generate well mimicked output, but to some of us it's so obvious and buggy that it's a waste of time.
      - ben_w 3y ago
        If it is convincing most people, then surely by definition it's closer to us than to a Markov chain.
HKH2 3y ago
I don't work in tech and I use LLaMa. It is great for writing overviews of a subject which I use to make sure I have most of my bases covered (and helping me see which of my ideas are more original), getting more examples, along with other creative tasks. Getting it to compare two subjects to see what they have in common is usually quite helpful.
- gumballindie 3y ago
  The issue is i tried using these types of tools for a while and each response is inaccurate. It becomes annoying to constantly ask it to correct it or having to doubt the output. Works well for what is statistically likely to be abundant in information but you know that’s usually of little value.
  - HKH2 3y ago
    I use them for brainstorming. I get such great ideas, I don't mind if they miss now and then. Information can be abundant yet hidden in obtuse or misleading writing. LLMs produce writing that is quite accessible.
MattRix 3y ago
I’m a little perplexed at why you feel the need to refer to it in reductionist terms like this? It seems to provide a lot of value considering that it has 100 million weekly active users. I regularly find myself using it a dozen times a day. I feel like the only reason someone wouldn’t find it useful is if they weren’t ever doing anything new.
- gumballindie 3y ago
  Loads of useless products have hundreds of millions of users. Threads, Twitter, Facebook, and so on. Just spamming forums to promote this product wont increase its utility.
  - MattRix 3y ago
    None of the things you mentioned are useless. Perhaps you‘re just being needlessly critical? Or is this just a form of trolling?
  - Turing_Machine 3y ago
    Why do you get make the decisions about what other people find useful?
ben_w 3y ago
It can be an infinitely patient teacher for physics and German, that never gets frustrated by me getting the conjugations wrong or having whacky ideas like getting lift from plasma held in a magnetic field. It can write boring code, like the OpenStreetMap query to find all supermarkets within 50x50km of Berlin. It can do code and literature reviews of my solo projects and stories. It can take a list of ingredients and suggest things I can cook with them, and unlike "real" recipe websites it can do this without telling me how their grandmother used to comfort them with this on a beautiful autumn weekend, when… [1000 words and four ad breaks later] … and then when the website actually gets to the recipe it turns out to be unsuitable anyway. It can write form letters It can help role-play difficult real-life scenarios. - I don't expect perfection in any of these things. If it was perfect, us software developers would already be redundant, as would all the lawyers, the accountants, and anyone else whose job can be reduced to images and words.