5 ms·

It'll decimate a lot of the service sector, that is white collar workers, and that includes us; it'll decimate it in the same way industrialisation decimated a

It'll decimate a lot of the service sector, that is white collar workers, and that includes us; it'll decimate it in the same way industrialisation decimated a lot of blue collar jobs; and control of a centralised AI will give companies and governments incredible power. But comparing it to electricity is crazy.

belter 4y ago
I just asked ChatGPT the definition of Confidence Interval in Statistics. It wrote a very elegant, authoritative and eloquent, but completely wrong explanation, that interestingly, matches the most common misconception around Confidence Intervals. Specialist jobs are not in danger yet. However the level of bullshitting will increase to such levels, there is a business opportunity here: One for tools capable of identifying the errors, spit out by these very impressive automatic parrots.
- Closi 4y ago
  The assumption in your post is that the ‘bullshitting’ won’t/can’t be solved within a few iterations. Given the decrease of this already in GPT 4, it’s entirely possible that this will be solved soon. We are still right at the birth of this technology - assuming this technology takes 30 years to fully mature it’s like we are in 1978 trying to predict the iPhone.
  - poulsbohemian 4y ago
    Actually - that’s a great example. If you go back to the 40s/50s/60s/70s/80s/90s - you’ll see all sorts of grand predictions about how technology will change our lives. And then 20 years go by with very gradual incremental changes. Anyone working in tech realizes that AI has been a work in progress for a long time, it will continue to make in-roads, but is it really going to put us out of work in the next 5-10 years?
    - thefounder 4y ago
      Well we still don't have flying cars or fancy spaceships not to mention immortality other SF predictions so I wouldn't hold my breath for a AGI-like discovery. Should I mention we don't have full self driving either...?
    - Closi 4y ago
      Gradual incremental changes until there is suddenly an explosion of change and adoption, if the PC and Internet are anything to go by. In my view it will probably change the way we work to a large extent in 10 years, and tech work will probably be entirely different in 30 years, but like you say these are impossible things to know and an intelligent person could entirely disagree.
  - Xeamek 4y ago
    >The assumption in your post is that the ‘bullshitting’ won’t/can’t be solved within a few iterations. Because it can't, at least not how chatGPT currently operates. Is it possible there will be more breakthroughs soon? Sure. But it's also possible that we just made a massive leap and we won't see any major revolutions any time soon >We are still right at the birth of this technology - assuming this technology takes 30 years to fully mature it’s like we are in 1978 trying to predict the iPhone. Except we don't know where we are. We might verry well be close to the peak, rather then beginning.
    - Closi 4y ago
      It’s unlikely we are close to the peak IMO - the rate of improvement and discoveries appears to be increasing rather than decreasing, implying that we are still at the start of the curve. Progress doesn’t usually immediately stop, you would expect it to slowly flatten. It will absolutely need to be combined with other breakthroughs - but this is already happening. There are architectures in the works that allow AI agents to perform accurate calculations, search the internet, and check knowledge graphs. Then there is a whole raft of technologies yet to emerge which will embed this into real world work flows, which hasn’t really started happening in a significant way yet (bits and bobs but not to where it will inevitably will be). I just can’t see a world where there isn’t more progress in these technologies - I don’t know if they will be a ‘good’ thing or not for society as we know it, however with the rate of change at the moment we are unlikely to be anywhere near the destination of where the train goes.
- tarkin2 4y ago
  One currently incorrect answer doesn't render the technology useless for all domains. You'll still need experts to check the output. But this will be their job--checking, refining, regenerating--rather than creating. And jobs underneath that level will disappear.
  - hedora 4y ago
    Haven’t those jobs already disappeared though? You’re basically describing the minimal skill set required of someone that uses a search engine at work.
    - tarkin2 4y ago
      You still have the write the code, write the report, etc, after looking at google. Now you won't need to.
hedora 4y ago
Why would this AI be centralized? You’ll probably be able to run the inference side on your phone soon. Training costs are only mid 7 figures and dropping exponentially.

belter 4y ago

I just asked ChatGPT the definition of Confidence Interval in Statistics. It wrote a very elegant, authoritative and eloquent, but completely wrong explanation, that interestingly, matches the most common misconception around Confidence Intervals. Specialist jobs are not in danger yet. However the level of bullshitting will increase to such levels, there is a business opportunity here: One for tools capable of identifying the errors, spit out by these very impressive automatic parrots.

Closi 4y ago
The assumption in your post is that the ‘bullshitting’ won’t/can’t be solved within a few iterations. Given the decrease of this already in GPT 4, it’s entirely possible that this will be solved soon. We are still right at the birth of this technology - assuming this technology takes 30 years to fully mature it’s like we are in 1978 trying to predict the iPhone.
- poulsbohemian 4y ago
  Actually - that’s a great example. If you go back to the 40s/50s/60s/70s/80s/90s - you’ll see all sorts of grand predictions about how technology will change our lives. And then 20 years go by with very gradual incremental changes. Anyone working in tech realizes that AI has been a work in progress for a long time, it will continue to make in-roads, but is it really going to put us out of work in the next 5-10 years?
  - thefounder 4y ago
    Well we still don't have flying cars or fancy spaceships not to mention immortality other SF predictions so I wouldn't hold my breath for a AGI-like discovery. Should I mention we don't have full self driving either...?
  - Closi 4y ago
    Gradual incremental changes until there is suddenly an explosion of change and adoption, if the PC and Internet are anything to go by. In my view it will probably change the way we work to a large extent in 10 years, and tech work will probably be entirely different in 30 years, but like you say these are impossible things to know and an intelligent person could entirely disagree.
- Xeamek 4y ago
  >The assumption in your post is that the ‘bullshitting’ won’t/can’t be solved within a few iterations. Because it can't, at least not how chatGPT currently operates. Is it possible there will be more breakthroughs soon? Sure. But it's also possible that we just made a massive leap and we won't see any major revolutions any time soon >We are still right at the birth of this technology - assuming this technology takes 30 years to fully mature it’s like we are in 1978 trying to predict the iPhone. Except we don't know where we are. We might verry well be close to the peak, rather then beginning.
  - Closi 4y ago
    It’s unlikely we are close to the peak IMO - the rate of improvement and discoveries appears to be increasing rather than decreasing, implying that we are still at the start of the curve. Progress doesn’t usually immediately stop, you would expect it to slowly flatten. It will absolutely need to be combined with other breakthroughs - but this is already happening. There are architectures in the works that allow AI agents to perform accurate calculations, search the internet, and check knowledge graphs. Then there is a whole raft of technologies yet to emerge which will embed this into real world work flows, which hasn’t really started happening in a significant way yet (bits and bobs but not to where it will inevitably will be). I just can’t see a world where there isn’t more progress in these technologies - I don’t know if they will be a ‘good’ thing or not for society as we know it, however with the rate of change at the moment we are unlikely to be anywhere near the destination of where the train goes.
tarkin2 4y ago
One currently incorrect answer doesn't render the technology useless for all domains. You'll still need experts to check the output. But this will be their job--checking, refining, regenerating--rather than creating. And jobs underneath that level will disappear.
- hedora 4y ago
  Haven’t those jobs already disappeared though? You’re basically describing the minimal skill set required of someone that uses a search engine at work.
  - tarkin2 4y ago
    You still have the write the code, write the report, etc, after looking at google. Now you won't need to.

hedora 4y ago

Why would this AI be centralized? You’ll probably be able to run the inference side on your phone soon. Training costs are only mid 7 figures and dropping exponentially.