3 ms·

The problem is you struggle to hack together a sub optimal solution because you don't have the resources to figure out anything better. It didn't make me better

The problem is you struggle to hack together a sub optimal solution because you don't have the resources to figure out anything better. It didn't make me better, it was a waste of time. Remember experts exchange? Uhg.. Today I can quickly be productive in unfamiliar domains, and learn while asking questions to an AI that is available 24/7, and and has an extremely deep knowledge of so many things. Personally even as an experienced programmer I have learned so much in the last year, greatly accelerated by AI. Kids are going to be better off for it, and with it will achieve incredible things.

aleph_minus_one 2y ago
> Today I can quickly be productive in unfamiliar domains Today, you can quickly get a shallow, superficial understanding of the domain that might impress some people who know barely anything about the respective domain. > and learn while asking questions to an AI that [...] has an extremely deep knowledge of so many things. ROFL
- bottlepalm 2y ago
  You're threatened by AI, I get that.
  - aleph_minus_one 2y ago
    I am rather threatened by people who fall for the AI hype. :-D
    - bottlepalm 2y ago
      Yes, often when threatened we revert to a state of denial. Try to trivialize and disarm the threat in our mind.
anthk 2y ago
>better Bullshit. I've seen blind idiots kids pasting Solaris commands into Linux servers, without having any clue of what were they doing. Kids are going to be really screwed when the AI output converges into more Markov-chains like bullshit as it's being self-feeding with it's own output, creating something like a big but pompous Megahal/Hailo clone. If any, we the older Millenials are trained to do hard tasks with just the manuals sitting on a table. The rest will be clueless.
- Philpax 2y ago
  > Kids are going to be really screwed when the AI output converges into more Markov-chains like bullshit as it's being self-feeding with it's own output, creating something like a big but pompous Megahal/Hailo clone. This isn't going to happen. The literature on model collapse suggests it occurs when your model is fed on majority synthetic data, which is not how anyone is training models. Even if they were, do you think they're going to ship something that performs noticeably worse than its predecessor?
  - anthk 2y ago
    You said it. Traning. But later, most of the input of data it's being inputted back from AI's output. Because the end users will use far more the output data from AI than feeding it from remote sources far from the original. Disasters will happen, just wait.
    - hattmall 2y ago
      > do you think they're going to ship something that performs noticeably worse than its predecessor? This happens all the time. Once the first gen devs are off, the releases tend to alternate. A decline from the existing version and then an improvement that mostly fixes what they made worse in the prior release. e.g. Windows XP -> Vista -> 7 -> 8 -> 10 The nature of AI though may make it more difficult to distinguish the failings immediately which could result in an irreversible inflection point.
      - anthk 2y ago
        AI's are not OSes. Once the source of training comes mostly from the AI itself instead of human curated sources, what will you get over time it's the same thing the Megahal/Hailo chatbots parroted a few decades ago.
  - Timwi 2y ago
    > do you think they're going to ship something that performs noticeably worse than its predecessor? Yes. Capitalist software vendors do that all the time. Google search has been getting continually worse over the decades. Windows is more bloated than ever. AI is going to be no different.
CalRobert 2y ago
Hell, I'm old and had been meaning to learn React for years but never really stuck with it (I have things to do). With GPT it was pretty easy. I still read the docs, but I don't get stuck for hours on syntax minutiae and the like.

aleph_minus_one 2y ago

> Today I can quickly be productive in unfamiliar domains Today, you can quickly get a shallow, superficial understanding of the domain that might impress some people who know barely anything about the respective domain. > and learn while asking questions to an AI that [...] has an extremely deep knowledge of so many things. ROFL

bottlepalm 2y ago
You're threatened by AI, I get that.
- aleph_minus_one 2y ago
  I am rather threatened by people who fall for the AI hype. :-D
  - bottlepalm 2y ago
    Yes, often when threatened we revert to a state of denial. Try to trivialize and disarm the threat in our mind.

anthk 2y ago

>better Bullshit. I've seen blind idiots kids pasting Solaris commands into Linux servers, without having any clue of what were they doing. Kids are going to be really screwed when the AI output converges into more Markov-chains like bullshit as it's being self-feeding with it's own output, creating something like a big but pompous Megahal/Hailo clone. If any, we the older Millenials are trained to do hard tasks with just the manuals sitting on a table. The rest will be clueless.

Philpax 2y ago
> Kids are going to be really screwed when the AI output converges into more Markov-chains like bullshit as it's being self-feeding with it's own output, creating something like a big but pompous Megahal/Hailo clone. This isn't going to happen. The literature on model collapse suggests it occurs when your model is fed on majority synthetic data, which is not how anyone is training models. Even if they were, do you think they're going to ship something that performs noticeably worse than its predecessor?
- anthk 2y ago
  You said it. Traning. But later, most of the input of data it's being inputted back from AI's output. Because the end users will use far more the output data from AI than feeding it from remote sources far from the original. Disasters will happen, just wait.
  - hattmall 2y ago
    > do you think they're going to ship something that performs noticeably worse than its predecessor? This happens all the time. Once the first gen devs are off, the releases tend to alternate. A decline from the existing version and then an improvement that mostly fixes what they made worse in the prior release. e.g. Windows XP -> Vista -> 7 -> 8 -> 10 The nature of AI though may make it more difficult to distinguish the failings immediately which could result in an irreversible inflection point.
    - anthk 2y ago
      AI's are not OSes. Once the source of training comes mostly from the AI itself instead of human curated sources, what will you get over time it's the same thing the Megahal/Hailo chatbots parroted a few decades ago.
- Timwi 2y ago
  > do you think they're going to ship something that performs noticeably worse than its predecessor? Yes. Capitalist software vendors do that all the time. Google search has been getting continually worse over the decades. Windows is more bloated than ever. AI is going to be no different.

CalRobert 2y ago

Hell, I'm old and had been meaning to learn React for years but never really stuck with it (I have things to do). With GPT it was pretty easy. I still read the docs, but I don't get stuck for hours on syntax minutiae and the like.