3 ms·

The main thing I want LLMs to help with is obscure stuff. If the answer is mainstream or well known, I can get it more reliably through a normal web search.

JohnFen 2y ago
By the very nature of its training data, the obscure stuff will be the stuff these models are the worst at.
- awfulneutral 2y ago
  Yeah, I guess I should have said I think AI chat interfaces in general would be more useful if they could help with the questions that weren't already easy to answer. Sometimes I can get use out of an LLM chat, but it's really hit or miss.
- sroussey 2y ago
  Depends on how obscure and how much SEO competition there is.
DrSiemer 2y ago
By far their most useful and reliable feature is asking them for stuff I already know, but can't be bothered to look up the details for or write out myself. It saves me from dealing with thick layers of SEO crap, cookie banners and many of the incorrect answers. Sure, it would be nice if we had an LLM that could help us with the more unusual questions. But for that we would either need much better training data, full of all those unknown answers (which does not exist), or a totally different system that can actually think for itself instead of just mimicking existing human output.
- awfulneutral 2y ago
  That's interesting, you mean something you forgot but could verify when you see the answer? Or something you know, but you need a write-up for it or something?
  - DrSiemer 2y ago
    Many of the work I do involves things I have done before in some form. But all that knowledge is spread out over multiple languages and libraries and the details are often hazy. LLM's are excellent at stuff that has been done before and they are also fantastic for transposing code from one language to the other. Looking up code in old projects takes time and brain RAM. Spelling out the steps in pseudo code and then prompt guiding an LLM through figuring out the required code often works great for smaller tasks. I've also learned many new tricks by just asking it why it made certain choices.
    - awfulneutral 2y ago
      I see, thanks for the explanation!

JohnFen 2y ago

By the very nature of its training data, the obscure stuff will be the stuff these models are the worst at.

awfulneutral 2y ago
Yeah, I guess I should have said I think AI chat interfaces in general would be more useful if they could help with the questions that weren't already easy to answer. Sometimes I can get use out of an LLM chat, but it's really hit or miss.
sroussey 2y ago
Depends on how obscure and how much SEO competition there is.

DrSiemer 2y ago

By far their most useful and reliable feature is asking them for stuff I already know, but can't be bothered to look up the details for or write out myself. It saves me from dealing with thick layers of SEO crap, cookie banners and many of the incorrect answers. Sure, it would be nice if we had an LLM that could help us with the more unusual questions. But for that we would either need much better training data, full of all those unknown answers (which does not exist), or a totally different system that can actually think for itself instead of just mimicking existing human output.

awfulneutral 2y ago
That's interesting, you mean something you forgot but could verify when you see the answer? Or something you know, but you need a write-up for it or something?
- DrSiemer 2y ago
  Many of the work I do involves things I have done before in some form. But all that knowledge is spread out over multiple languages and libraries and the details are often hazy. LLM's are excellent at stuff that has been done before and they are also fantastic for transposing code from one language to the other. Looking up code in old projects takes time and brain RAM. Spelling out the steps in pseudo code and then prompt guiding an LLM through figuring out the required code often works great for smaller tasks. I've also learned many new tricks by just asking it why it made certain choices.
  - awfulneutral 2y ago
    I see, thanks for the explanation!