Comment by cat_plus_plus
18 days ago
I have no idea what everyone is talking about. LLMs are based on relatively simple math, inference is much easier to learn and customize than say Android APIs. Once you do you can apply familiar programming style logic to messy concepts like language and images. Give you model a JSON schema like "warp_factor": Integer if you don't want chatter, that's way better than Star Trek computer could do. Or have it write you a simple domain specific library on top of Android API that you can then program from memory like old style BASIC rather than having to run to stack overflow for evwery new task.
You can’t reason about inference (or training) of LLMs on the semantic level. You can’t predict the output of an LLM for a specific input other than by running it. If you want the output to be different in a specific way, you can’t reason with precision that a particular modification of the input, or of the weights, will achieve the desired change (and only that change) in the output. Instead, it’s like a slot machine that you just have to try running again.
The fact that LLMs are based on a network of simple matrix multiplications doesn’t change that. That’s like saying that the human brain is based on simple physical field equations, and therefore its behavior is easy to understand.
That’s like saying that the human brain is based on simple physical field equations, and therefore its behavior is easy to understand.
Right, which is the point: LLMs are much more like human coworkers than compilers in terms of how you interact with them. Nobody would say that there's no point to working with other people because you can't predict their behavior exactly.
This thread is about what software developers like. It’s common knowledge that many programmers like working with computers because that’s different in specific ways from working with people. So saying that LLMs are just like people doesn’t help here.
7 replies →
> LLMs are much more like human coworkers than compilers in terms of how you interact with them.
Human coworkers are much more predictable. A workplace where people act similarly to LLM would be a complete zoo. Imagine asking for an endpoint modification and the result is a broken backend. Or brainstorming with a PM and the reply are "you're absolutely right, whatever I was saying was completely wrong, but let me repeat it in a different manner".
1 reply →
Nobody would say:
"...there's no point to working with other people because you can't predict their behavior exactly."
Because you CAN predict coworker behavior to a useful point. Ex, they'll probably reply to that email on Monday. They'll probably show you a video that you find less amusing than they do.
With LLMs you can't be quite sure whether they will make something up, forget a key detail, hide a mistake that will obviously be found out when everything breaks, etc. Stupid things that most employable people wouldn't do, like building a car and forgetting the wheels.
> LLMs are much more like human coworkers
Specifically they are like Julius, the colleague managers like but is a drag on everyone else.
https://ploum.net/2024-12-23-julius-en.html
What are you inputs and outputs? If inputs are zip files and outputs is uncompressed text, don't use an LLM. If inputs are English strings and outputs are localized strings, LLMs are way more accurate than any procedural code you might attempt for the purpose. Plus changing the style of outputs by modifying inputs/weights is also easier, you just need to provide a few thousand samples rather than think of every case. Super relevant for human coding, how many hobbyists or small businesses have teams of linguists on staff?