5 ms·

Not at all. There is very little, practically zero of my day to day AI can do. The other threat is of course reducing the number of people needed. Like an auto

Not at all. There is very little, practically zero of my day to day AI can do. The other threat is of course reducing the number of people needed. Like an automatic loom. Again not to worried. I think that will generate new demand. The deepfake side of things is more scary IMO.

noname120 2y ago
What is something knowledge workers will still do better than AI in 5 years? Automatic loom is something different because — like other progresses in history — it just automated manual work. This and other industrial progresses gradually led most jobs to shift from manual to office-type work. Computers automated the boring/“mechanical” parts of thinking (simulations, solving equations, etc.) but now they are taking over the whole thinking part: intuition, systemization, requirements gathering, soft skills, etc. Nothing will be spared as far as I see. Stakeholders at the top of the ladder will still be there (as they will own the means of production — AI + compute centers) but they will mostly just instruct AI to do what they want. AI will ask clarifying questions, do research, maybe spin up a few dozen of instances of themselves to simulate the emergent results of having a group of people/an organization design and think things through, rather than a lone-wolf kind of thing.
- aqueueaqueue 2y ago
  What you say is certainly possible but I have not seen anything from LLMs so far that make me think it is probable. They just make too many mistakes and are not really thinking. I think they are a dead end. If the scenario you talk about happens it is because we discover some new architecture previously unthought of as well as advancement in processor efficiency and price.
  - noname120 2y ago
    Consensus in the machine learning community is that current architectures are enough for AGI, no new architecture needed. There is plenty of leeway as well into how much money we can pour into training and if the linear scaling law for model capabilities still holds — and we have no reason to believe yet that they won't we have the resources (both financial and technological) to reach this within 10-20 years tops.
    - pillefitz 2y ago
      A few counter points: Most developers prefer Claude 3.5 (or the slightly improved 3.7 since a few days back), which is a year old already. What was supposedly ChatGPT 5 will be called 4.5 instead, after two multi-month and hugely expensive training runs. Grok 3 was trained on a massive cluster of 200.000 H100 and still not beating SOTA models. There's certainly not "a clear way forward", even though one may hope the billions invested will lead to algorithmic breakthroughs.
    - JohnFen 2y ago
      > Consensus in the machine learning community is that current architectures are enough for AGI, no new architecture needed. I don't think there's anything like consensus on this. There isn't even consensus on whether AGI is possible in the first place.
      - unmole 2y ago
        There isn't even consensus on what AGI means.

noname120 2y ago

What is something knowledge workers will still do better than AI in 5 years? Automatic loom is something different because — like other progresses in history — it just automated manual work. This and other industrial progresses gradually led most jobs to shift from manual to office-type work. Computers automated the boring/“mechanical” parts of thinking (simulations, solving equations, etc.) but now they are taking over the whole thinking part: intuition, systemization, requirements gathering, soft skills, etc. Nothing will be spared as far as I see. Stakeholders at the top of the ladder will still be there (as they will own the means of production — AI + compute centers) but they will mostly just instruct AI to do what they want. AI will ask clarifying questions, do research, maybe spin up a few dozen of instances of themselves to simulate the emergent results of having a group of people/an organization design and think things through, rather than a lone-wolf kind of thing.

aqueueaqueue 2y ago
What you say is certainly possible but I have not seen anything from LLMs so far that make me think it is probable. They just make too many mistakes and are not really thinking. I think they are a dead end. If the scenario you talk about happens it is because we discover some new architecture previously unthought of as well as advancement in processor efficiency and price.
- noname120 2y ago
  Consensus in the machine learning community is that current architectures are enough for AGI, no new architecture needed. There is plenty of leeway as well into how much money we can pour into training and if the linear scaling law for model capabilities still holds — and we have no reason to believe yet that they won't we have the resources (both financial and technological) to reach this within 10-20 years tops.
  - pillefitz 2y ago
    A few counter points: Most developers prefer Claude 3.5 (or the slightly improved 3.7 since a few days back), which is a year old already. What was supposedly ChatGPT 5 will be called 4.5 instead, after two multi-month and hugely expensive training runs. Grok 3 was trained on a massive cluster of 200.000 H100 and still not beating SOTA models. There's certainly not "a clear way forward", even though one may hope the billions invested will lead to algorithmic breakthroughs.
  - JohnFen 2y ago
    > Consensus in the machine learning community is that current architectures are enough for AGI, no new architecture needed. I don't think there's anything like consensus on this. There isn't even consensus on whether AGI is possible in the first place.
    - unmole 2y ago
      There isn't even consensus on what AGI means.