6 ms·

Same. My friends in other white-collar-ish jobs, which require lots of writing, are smitten with ChatGPT and think it's amazing. Friends in tech are largely dis

Same. My friends in other white-collar-ish jobs, which require lots of writing, are smitten with ChatGPT and think it's amazing. Friends in tech are largely dismissive/critical of it's abilities while being skeptical/fearful of its impact. My take is that, for most non-tech people, this is their first experience directing a computer to perform a precise task - ie programming. They're accustomed to using applications, not having a hand in making them. Maybe they've used Excel before and felt a bit of this power. But ChatGPT allows them to dream up a novel idea and get the computer to execute it - something that feels like magic to non-tech folks but is rather pedestrian for most of us in tech.

voidfunc 3y ago
This reason makes sense, but that's insanely powerful (and valuable) if it ends up working out. It basically democratizes programming.
- butler14 3y ago
  That's exactly what it does. It closes the gap. Even now with all its shortcomings and issues. It's actively solving problems for people and saving time / creating efficiency. That has value and economic utility. I honestly don't get what a lot of the cynics in this thread and their "highly technical/IT friends" are missing.
  - perrygeo 3y ago
    It's not saving time or creating efficiency for everyone equally. It's absolutely democratizing some complex tasks. I'll stick to software here, where AI enhances the abilities of non-programmers to give them a taste of what programmers have been doing for decades. But does it enhance the end product? Does it improve upon the work of already competent developers? Does it actually solve the real problems that software engineers face? Highly debatable. It certainly makes cranking out code more "efficient" - but anyone who's every created software knows that cranking out lines of code is a terrible metric for success. Poor quality code has less than zero value, it's a liability. As the prevalence of bot-generated code goes up, it will place an increasingly high burden on actual professionals to clean up the mess.
- byteware 3y ago
  how is programming "democratized"?
  - square_usual 3y ago
    More people can now cobble barely functioning python scripts together and shoot themselves in the foot when it does something they don't understand :>
    - throwuwu 3y ago
      The point is they don’t need any code. They can copy paste a csv or something even messier and ask it to write a somewhat personalized email to each person in the list, if they’re smart about it they can include a few small details with each input entry that get included in the email. They also now have access to most NLP tools like sentiment analysis or feature extraction which allows them to process large amounts of text and extract valuable insights. They don’t need Python or a programmer as long as they’re smart enough to try out new things and learn the ins and outs of these models.
      - square_usual 3y ago
        Realistically, nobody is working with CSVs so small that they fit within ChatGPT's window. I've helped friends and faily use ChatGPT to generate a python script that did what they wanted with CSV, and they all started with attempting to paste a ~x00 line CSV in the chat.
- SoftTalker 3y ago
  Except the essential difficulty with programming is not typing the code or even understanding the syntax and idioms of a particular language.
ozim 3y ago
I did not see a way where you can tell AI “make me a crud app for storing database of my post stamp collection” and it executes that. It is quite a stretch to say that lay people can effortlessly write application only with AI whatever it is copilot or chatgpt. You still have to know a lot of things to build simple crud to store data about stamp collection. LLMs are not changing that.
- perrygeo 3y ago
  LLMs are not changing that, yet. But well over a decade ago, we had high-level frameworks for translating "models" into all the gritty details of a crud app. Model description goes in. API endpoints, database migrations, admin UI, all auto-generated. I even developed something like it based on Django 0.96 (now abandoned). It's not a stretch of the imagination to generate your model.py with model.txt containing the problem description in plain text.
  - SoftTalker 3y ago
    Do you know any real professional software developers who actually use these modeling tools of their own volition and not because some manager fell for a golf-and-steak-dinner sales pitch from the vendor? My experience with every one of them is that getting the model to the point where it will generate something useful is more work than just writing the useful part myself.
    - perrygeo 3y ago
      As I alluded to in my comment, I have a ton of experience with such systems. And no, it was not some "golf-and-steak-dinner" - it was funded by a non-profit environmental organization, literally. I'm serious, AMA. Don't make shit up - reconsider your first point. Your second point re: how model-driven code generation can be a dead-end... I totally agree. I didn't say it was a good idea! But it was motivated by a real need: generating lots of web apps in a particular domain.

voidfunc 3y ago

This reason makes sense, but that's insanely powerful (and valuable) if it ends up working out. It basically democratizes programming.

butler14 3y ago
That's exactly what it does. It closes the gap. Even now with all its shortcomings and issues. It's actively solving problems for people and saving time / creating efficiency. That has value and economic utility. I honestly don't get what a lot of the cynics in this thread and their "highly technical/IT friends" are missing.
- perrygeo 3y ago
  It's not saving time or creating efficiency for everyone equally. It's absolutely democratizing some complex tasks. I'll stick to software here, where AI enhances the abilities of non-programmers to give them a taste of what programmers have been doing for decades. But does it enhance the end product? Does it improve upon the work of already competent developers? Does it actually solve the real problems that software engineers face? Highly debatable. It certainly makes cranking out code more "efficient" - but anyone who's every created software knows that cranking out lines of code is a terrible metric for success. Poor quality code has less than zero value, it's a liability. As the prevalence of bot-generated code goes up, it will place an increasingly high burden on actual professionals to clean up the mess.
byteware 3y ago
how is programming "democratized"?
- square_usual 3y ago
  More people can now cobble barely functioning python scripts together and shoot themselves in the foot when it does something they don't understand :>
  - throwuwu 3y ago
    The point is they don’t need any code. They can copy paste a csv or something even messier and ask it to write a somewhat personalized email to each person in the list, if they’re smart about it they can include a few small details with each input entry that get included in the email. They also now have access to most NLP tools like sentiment analysis or feature extraction which allows them to process large amounts of text and extract valuable insights. They don’t need Python or a programmer as long as they’re smart enough to try out new things and learn the ins and outs of these models.
    - square_usual 3y ago
      Realistically, nobody is working with CSVs so small that they fit within ChatGPT's window. I've helped friends and faily use ChatGPT to generate a python script that did what they wanted with CSV, and they all started with attempting to paste a ~x00 line CSV in the chat.
SoftTalker 3y ago
Except the essential difficulty with programming is not typing the code or even understanding the syntax and idioms of a particular language.

ozim 3y ago

I did not see a way where you can tell AI “make me a crud app for storing database of my post stamp collection” and it executes that. It is quite a stretch to say that lay people can effortlessly write application only with AI whatever it is copilot or chatgpt. You still have to know a lot of things to build simple crud to store data about stamp collection. LLMs are not changing that.

perrygeo 3y ago
LLMs are not changing that, yet. But well over a decade ago, we had high-level frameworks for translating "models" into all the gritty details of a crud app. Model description goes in. API endpoints, database migrations, admin UI, all auto-generated. I even developed something like it based on Django 0.96 (now abandoned). It's not a stretch of the imagination to generate your model.py with model.txt containing the problem description in plain text.
- SoftTalker 3y ago
  Do you know any real professional software developers who actually use these modeling tools of their own volition and not because some manager fell for a golf-and-steak-dinner sales pitch from the vendor? My experience with every one of them is that getting the model to the point where it will generate something useful is more work than just writing the useful part myself.
  - perrygeo 3y ago
    As I alluded to in my comment, I have a ton of experience with such systems. And no, it was not some "golf-and-steak-dinner" - it was funded by a non-profit environmental organization, literally. I'm serious, AMA. Don't make shit up - reconsider your first point. Your second point re: how model-driven code generation can be a dead-end... I totally agree. I didn't say it was a good idea! But it was motivated by a real need: generating lots of web apps in a particular domain.