4 ms·

There was a moment when google introduced autocomplete and it was a game changer. LLMs are still waiting for their autocomplete moment: when they become an ext

by seydor 2y ago

There was a moment when google introduced autocomplete and it was a game changer. LLMs are still waiting for their autocomplete moment: when they become an extension of the keyboard and complete our thoughts so fast, that i could write this article in 2 minutes. That will feel magical. The speed is currently missing

GiorgioG 2y ago
No, that is not the only thing missing. Right now AI continues to make mistakes a child learns not to make by age 10…don’t make shit up. And by time you’re an adult you figure out how to manage not to forget (by whatever means necessary) the first thing your boss told you to do after he added a new ask.
- roflmaostc 2y ago
  That's not entirely true. AI can solve math puzzles better than 99.9% of population. Yes AI makes mistakes, so do humans very often.
  - root_axis 2y ago
    Humans make mistakes, sure, but if a human starts hallucinating we immediately lose trust in them. > AI can solve math puzzles better than 99.9% of population So can a calculator.
  - skydhash 2y ago
    > AI can solve math puzzles better than 99.9% of population I've studied electronic engineering and then switched to software engineering as a career, and I can say the only time I've been exposed to math puzzles were in academic settings. The knowledge is nice and help with certain problem solving, but you can be pretty sure I will reach out to a textbook and a calculator before trying to brute-force one such puzzle. The most important thing in my daily life is understand the given task, do it correctly, and report about what I've done.
    - SoftTalker 2y ago
      Yep this exactly. If I ever feel that I'm solving a puzzle at work, I stop and ask for more information. Once my task is clear, I start working on it. Learned long ago that puzzle solving is almost always solving the wrong problem and wasting time.
      - skydhash 2y ago
        Maybe that's where the disconnect comes from. For me understanding comes before doing. And coding for me is doing. There may be cases when I assumes that my understanding was complete and the feedback (errors during compiling and testing) told me I'm wrong, but that just triggers another round of research to seek understanding. Puzzle solving is only for when information are not available (reverse engineering, closed systems,...) but there's a lot of information out there for the majority of tasks. I'm amazed when people spend hours trying to vibe code something, where they could spend just a few minutes reading about the system and comes up with a working solution (or find something that already works).
  - masswerk 2y ago
    > so do humans very often. While I do see this argument made quite frequently, doesn't any professional effort center in procedures employed particularly to avoid mistakes? Isn't this really the point of professional work (including professional liabilities)?
MichaelZuo 2y ago
But who will actually read the “autocompleted” text? At that point any other human being will likely also have one to scan incoming text.
kadushka 2y ago
For me, it’s the opposite – speed is there, intelligence is still lacking sometimes. I’m OK waiting 10 minutes with o1-pro, but I want a deep insight into the issue I’m brainstorming. Hopefully GPT-5 will deliver.
roflmaostc 2y ago
For me the speed is already there. LLMs can write boilerplate code at least an order of magnitude faster than I can. Just today I generate U-Net code for a certain scenario. I had to tweak some parameters, at the end I got it working in <1hr.
- diggan 2y ago
  > LLMs can write boilerplate code at least an order of magnitude faster than I can. This is my biggest fear with everyone adopting LLMs without considering the consequences. In the past, I used "Do I have to write a lot of boilerplate here?" as a sort of litmus test for figuring out when to refactor. If I spend the entire day just writing boilerplate, I'm 99% sure I'm doing the wrong thing, at least most of the time. But now, junior developers won't even get the intuition that if they're spending the entire day just typing boilerplate something is wrong, instead they'll just get the LLM to do it and there is no careful thoughts/reflections about the design and architecture. Of course, reflection is still possible, but I'm afraid it won't be as natural and "in your face" which kind of forces you to learn it, instead it'll only be a thing for people who consider it in the first place.
  - warkdarrior 2y ago
    The question to ask yourself is what value one gets from that refactor. Is the software faster? Has more functionality? Cheaper to operate? Reduce time to market. These would be benefits to the user, and I'd venture to say the refactor does not impact the user. The refactor will impact the developer. Maybe the code is now more maintainable, or easier to integrate, or easier to test. But this is where I expect LLMs will make a lot of progress - - they will not need clean, well structured code. So the refactor, in the long run, is not useful to a developer with an LLM sidekick.
  - YurgenJurgensen 2y ago
    Or: An important thing that was taught to me on the first day of my C++ class was ‘no project was ever late because the typing took too long’.
  - jamauro 2y ago
diggan 2y ago
> The speed is currently missing I feel like the opposite/something else is missing. I can write lots of text quickly, if I just write down my unfiltered stream of thoughts, both together with and without an LLM, but what's the point? What takes really long time, is making a large text contain just the important parts. Saying less takes longer time, at least for me, but hopefully saves the time and effort for people reading it. At least that's the idea. The quote "If I had more time, I would have written a shorter letter" comes to mind.
- johnisgood 2y ago
  Unfiltered stream of thoughts: I wrote a lengthy note that is quite incoherent and disorganized, and I intend to use the LLM to organize it. Ugh.
WA 2y ago
You can do this in Cursor already. Write a README or a SPEC file and the LLM will try to "complete your thoughts", except that it's off often enough. I find this hugely distracting, as if someone always interrupts your train of thought with random stuff.
johnisgood 2y ago
Great idea! And scary.
SoftTalker 2y ago
> There was a moment when google introduced autocomplete and it was a game changer. I remember.... I turned it off immediately.
leftbit 2y ago
But why should I put my time into reading and thinking about your article if you didn't think it worth your time to actually think about and write it? Hope the Next Big Thing (TM) is the Electric Monk.

by seydor 2y ago

GiorgioG 2y ago

No, that is not the only thing missing. Right now AI continues to make mistakes a child learns not to make by age 10…don’t make shit up. And by time you’re an adult you figure out how to manage not to forget (by whatever means necessary) the first thing your boss told you to do after he added a new ask.

roflmaostc 2y ago
That's not entirely true. AI can solve math puzzles better than 99.9% of population. Yes AI makes mistakes, so do humans very often.
- root_axis 2y ago
  Humans make mistakes, sure, but if a human starts hallucinating we immediately lose trust in them. > AI can solve math puzzles better than 99.9% of population So can a calculator.
- skydhash 2y ago
  > AI can solve math puzzles better than 99.9% of population I've studied electronic engineering and then switched to software engineering as a career, and I can say the only time I've been exposed to math puzzles were in academic settings. The knowledge is nice and help with certain problem solving, but you can be pretty sure I will reach out to a textbook and a calculator before trying to brute-force one such puzzle. The most important thing in my daily life is understand the given task, do it correctly, and report about what I've done.
  - SoftTalker 2y ago
    Yep this exactly. If I ever feel that I'm solving a puzzle at work, I stop and ask for more information. Once my task is clear, I start working on it. Learned long ago that puzzle solving is almost always solving the wrong problem and wasting time.
    - skydhash 2y ago
      Maybe that's where the disconnect comes from. For me understanding comes before doing. And coding for me is doing. There may be cases when I assumes that my understanding was complete and the feedback (errors during compiling and testing) told me I'm wrong, but that just triggers another round of research to seek understanding. Puzzle solving is only for when information are not available (reverse engineering, closed systems,...) but there's a lot of information out there for the majority of tasks. I'm amazed when people spend hours trying to vibe code something, where they could spend just a few minutes reading about the system and comes up with a working solution (or find something that already works).
- masswerk 2y ago
  > so do humans very often. While I do see this argument made quite frequently, doesn't any professional effort center in procedures employed particularly to avoid mistakes? Isn't this really the point of professional work (including professional liabilities)?

MichaelZuo 2y ago

But who will actually read the “autocompleted” text? At that point any other human being will likely also have one to scan incoming text.

kadushka 2y ago

For me, it’s the opposite – speed is there, intelligence is still lacking sometimes. I’m OK waiting 10 minutes with o1-pro, but I want a deep insight into the issue I’m brainstorming. Hopefully GPT-5 will deliver.

roflmaostc 2y ago

For me the speed is already there. LLMs can write boilerplate code at least an order of magnitude faster than I can. Just today I generate U-Net code for a certain scenario. I had to tweak some parameters, at the end I got it working in <1hr.

diggan 2y ago
> LLMs can write boilerplate code at least an order of magnitude faster than I can. This is my biggest fear with everyone adopting LLMs without considering the consequences. In the past, I used "Do I have to write a lot of boilerplate here?" as a sort of litmus test for figuring out when to refactor. If I spend the entire day just writing boilerplate, I'm 99% sure I'm doing the wrong thing, at least most of the time. But now, junior developers won't even get the intuition that if they're spending the entire day just typing boilerplate something is wrong, instead they'll just get the LLM to do it and there is no careful thoughts/reflections about the design and architecture. Of course, reflection is still possible, but I'm afraid it won't be as natural and "in your face" which kind of forces you to learn it, instead it'll only be a thing for people who consider it in the first place.
- warkdarrior 2y ago
  The question to ask yourself is what value one gets from that refactor. Is the software faster? Has more functionality? Cheaper to operate? Reduce time to market. These would be benefits to the user, and I'd venture to say the refactor does not impact the user. The refactor will impact the developer. Maybe the code is now more maintainable, or easier to integrate, or easier to test. But this is where I expect LLMs will make a lot of progress - - they will not need clean, well structured code. So the refactor, in the long run, is not useful to a developer with an LLM sidekick.
- YurgenJurgensen 2y ago
  Or: An important thing that was taught to me on the first day of my C++ class was ‘no project was ever late because the typing took too long’.
- jamauro 2y ago

diggan 2y ago

> The speed is currently missing I feel like the opposite/something else is missing. I can write lots of text quickly, if I just write down my unfiltered stream of thoughts, both together with and without an LLM, but what's the point? What takes really long time, is making a large text contain just the important parts. Saying less takes longer time, at least for me, but hopefully saves the time and effort for people reading it. At least that's the idea. The quote "If I had more time, I would have written a shorter letter" comes to mind.

johnisgood 2y ago
Unfiltered stream of thoughts: I wrote a lengthy note that is quite incoherent and disorganized, and I intend to use the LLM to organize it. Ugh.

WA 2y ago

You can do this in Cursor already. Write a README or a SPEC file and the LLM will try to "complete your thoughts", except that it's off often enough. I find this hugely distracting, as if someone always interrupts your train of thought with random stuff.

johnisgood 2y ago

Great idea! And scary.

SoftTalker 2y ago

> There was a moment when google introduced autocomplete and it was a game changer. I remember.... I turned it off immediately.

leftbit 2y ago

But why should I put my time into reading and thinking about your article if you didn't think it worth your time to actually think about and write it? Hope the Next Big Thing (TM) is the Electric Monk.