4 ms·

Some of the comments mention LLMs. And they talked about LLMs having inaccuracies. OpenAI recently announced the ability to create custom versions of ChatGPT t

Some of the comments mention LLMs. And they talked about LLMs having inaccuracies. OpenAI recently announced the ability to create custom versions of ChatGPT that combine instructions, extra knowledge, and skills. https://openai.com/blog/introducing-gpts https://openai.com/blog/introducing-gpts I could see some of the mentioned expert communities start to build on top of that to offer newcomers a customized chat with domain knowledge.

zdw 3y ago
If this parent post is the quality of LLM generated content, the forum will go nowhere - it's not adding anything to the conversation.
- codetrotter 3y ago
  The point is that the problem is this: Experts want to provide material for newcomers. Newcomers don’t want to read the material. Using specially crafted material by the experts with an LLM can provide a way of delivering the expert knowledge to the newcomers in a format they can readily consume. Thus keeping the newcomers from flooding the forums with the same old basic questions over and over again.
  - zdw 3y ago
    Your argument seems to be that newcomers would rather use a interactive chat system than search the archives. As LLMs can hallucinate bad and false answers that often skip nuance, this would not be helpful in many cases, and possibly actively harmful. You're solving for experts not being bothered, but not for new folks being helped. A big part of learning is learning how to learn. Trying to skip that step with lying "AI" is fools gold.
  - Exoristos 3y ago
    There's no way I'm tediously interrogating some wordy half-witted chatbot when I could search the forums.
    - vasco 3y ago
      If when you arrive at a forum you use the search, the LLM isn't for you, and you're in the top 1% forum users.
  - therein 3y ago
    Back in the day and even today forums will throttle and captcha the search endpoint because it is ripe for abuse and computationally intensive. But nobody bats an eye as we throw clouds of GPUs to provide a chatbot experience.

zdw 3y ago

If this parent post is the quality of LLM generated content, the forum will go nowhere - it's not adding anything to the conversation.

codetrotter 3y ago
The point is that the problem is this: Experts want to provide material for newcomers. Newcomers don’t want to read the material. Using specially crafted material by the experts with an LLM can provide a way of delivering the expert knowledge to the newcomers in a format they can readily consume. Thus keeping the newcomers from flooding the forums with the same old basic questions over and over again.
- zdw 3y ago
  Your argument seems to be that newcomers would rather use a interactive chat system than search the archives. As LLMs can hallucinate bad and false answers that often skip nuance, this would not be helpful in many cases, and possibly actively harmful. You're solving for experts not being bothered, but not for new folks being helped. A big part of learning is learning how to learn. Trying to skip that step with lying "AI" is fools gold.
- Exoristos 3y ago
  There's no way I'm tediously interrogating some wordy half-witted chatbot when I could search the forums.
  - vasco 3y ago
    If when you arrive at a forum you use the search, the LLM isn't for you, and you're in the top 1% forum users.
- therein 3y ago
  Back in the day and even today forums will throttle and captcha the search endpoint because it is ripe for abuse and computationally intensive. But nobody bats an eye as we throw clouds of GPUs to provide a chatbot experience.