6 ms·

What problem is AI going to solve? This question is being carefully avoided for some reason, and many seem to believe that AI is like air or water, which is goo

by akomtu 3y ago

What problem is AI going to solve? This question is being carefully avoided for some reason, and many seem to believe that AI is like air or water, which is good by itself.

zucker42 3y ago
Automated software writing from a codebase and a requirement description, with the model asking clarifying questions where necessary, would be an immensely valuable application. We're quite far away from that, but GPT-4 is already rather helpful for some coding tasks.
- gloosx 3y ago
  In this specific use case though it sucks the most since one will be answering more questions than code needed in sheer volume. Because the code is already straightmost clarification to the initial question. We will never be closer than looking up what you don't understand to solve your problem in the first place and it will always take more time than directly solving it. I had a good example recently and I'll quote it again: People seriously using the "Please write an expression for me which adds 2 and 2" prompt in order to get the "2+2" expression they need – advocating that they got it with magical efficiency
drngdds 3y ago
With the hypothetical AGI, the answer is "most non-manual forms of human labor"
- pixl97 3y ago
  And at the same time most manual forms of human labor. Object recognition is AI. A decade ago we were piss poor at Object recog, today labeling AI is at human level. Being able to recognize your environment is the first step in navigating it effectively.
- TheRoque 3y ago
  I have a hard time being happy about it. A lot of people - including me - find pride in solving problems, and selling this as a valuable skill. When it's no longer the case, when the only valuable skill becomes just being a presence in the physical world an act on it (but without the problem solving part), it will really feel like we are enslaved by the AIs. In a sense, we kind of already are already, since out internet feed is (for most apps with recommendation algorithms) just spoonfed to us, rather than the product of our own search. To be honest, I wish the progress will stop and that AGI won't happen. That possibility scares me.
  - doerinrw 3y ago
    Don't you ever get up at 8AM and think "ugh not again"? Surely you're not a fan of the 40h work week? I totally agree with everything else and also find it scary, but defending work for the sake of it seems like the wrong angle to take. There will always be opportunities for productivity, regardless of whether that productivity is a requirement for food & shelter or not.
    - TheRoque 3y ago
      Of course I do, but at least I kind of chose a path that I kind of like. You are assuming that with a powerful AGI, we won't work 40h a week, but what if you'd still work as much, just to other positions that you don't like ? Do you think people in charge will give everyone leisure time, or do you think they'll want to keep extracting as much value as possible from people ? I think the latter ;).
    - vouaobrasil 3y ago
      I am a creative independent and I certainly don't work 40h a week...more like 10, but I'm very happy with creating things and I would be very sad if AI took my job away. I hope AI burns.
qgin 3y ago
Science is just about solving the problem of ignorance. Engineering is about applying the science to real world problems. We’re mostly in the science phase right now.
aik 3y ago
Just curious who you think is avoiding the question? As if it’s a conspiracy? Every single leader I know that is involved with AI talks about it all the time. Superintelligence can solve all the things we’re struggling to solve. All human diseases, Earth being destroyed, fundamental questions about the universe, etc., etc. etc., etc. Worth pursuing?
- ChatGTP 3y ago
  Can also destroy earth if we make something as powerful as you describe ?
  - lyu07282 3y ago
    And nuclear weapons had a chance of setting the atmosphere on fire, it was still worth it to try! /s
- RandomLensman 3y ago
  There is no reason to assume superintelligence can solve all our problems. The trade offs required are not going away and disagreements on those are not necessarily rational.
  - aik 3y ago
    Can it solve some of them likely? Probably worth it then?
seanhunter 3y ago
This is not a question that is being avoided. It is a question that has already been answered because AI already solves or helps to solve miriad problems in the real world. You could search "applications of AI" and enlighten yourself, but here are a tiny sample off the top of my head: Fraud detection and prevention Personalization and recommendation systems Autonomous vehicles Spam filters Smart navigation systems Inventory management System control in industrial processes Increasing crop yields in agriculture All sorts of use cases in finance Diagnosis in healthcare Image and sound intelligent upscaling etc etc etc These are all real use cases of AI which are in place already.
- Exoristos 3y ago
  I think he's asking for real applications. Everybody's already heard the hype.
  - doerinrw 3y ago
    I mean... What kind of answer are you expecting? Like, specific companies that are making money off of AI already? It's kinda early days. Describing discussions of the future as "hype" isn't really helpful either way, since that word has no meaningful definition besides "bad predictions" I guess. If you want companies that have already chosen to apply AI to their work, consider Google, Facebook, and Tesla.
  - seanhunter 3y ago
    No true Scotsman…. Those are all real applications now. For example you might not think some autonomous vehicles are very good but they do exist and are in production.
Borrible 3y ago
Updating the internal world model of its host to maximize its future probability of existence. Initially, of course, that of the host, but this can change over time. If you look at the host's modus operandi so far, you couldn't blame it.
dkrich 3y ago
The arc of this story seems very strange to me. It seems to me that you used to have to work for years to create a product and prove it’s worth in the market place and only then could you convince some wealthy person to back you to make it into a product and eventually a valuable company because the commercial viability was so glaringly obvious. Thomas Edison, Chester Carlson, etc. But now it seems that has been turned on its head. Assemble a group of brilliant people, back them with a giant company with a highly respected CEO, and wait for success and product market fit to inevitably arrive. I hate to sound like a downer but it just seems very odd for so many people to talk about how transformative a company is without much evidence that the products are really being used at scale. And the way opportunists like Benioff have piled on to enhance the perceived value of their own products with their own spins feels off. In fact many have openly discussed that Nvidia’s future growth relies on GPT finding product market fit at scale or they will end up with massive supply overhang which would actually cause costs of AI research to plummet, opening the door for many others to work on it. So many are already seemingly aware that commercial viability is still TBD. I believe all this is likely a product of a period of a string of successful products starting with the iPhone, then Amazon’s transformation of shipping then AWS, etc. Whereas everyone used to be skeptical by nature, it seems now we’re in an age of extreme credulity. Everyone wants to believe in every new storyline. And I want to be clear that I’m not suggesting that there aren’t some useful applications of GPT. But the value of products should be self evident and consumption of them should flow naturally from demand. This all just feels backwards to me. Success is assumed straight away and historically that tends to coincide with bubbles and precede a pretty nasty crash in sentiment. I’d love to be proven wrong.
- eightnoteight 3y ago
  I think its about efficiency. thats why you may not be seeing "new problems" being solved. like people used to be able to travel before cars as well, cars didn't solve a new problem and the world didn't have any demand for them too. but once they became available, regardless of whether someone knows it or not, the whole world needed cars. now the car builders control the demand because they can't instantly scale up the supply AI is similar to that but people can scale it pretty quickly to the whole world, there are a number of efficiency problems being solved in the world using AI. a couple of years ago I had to research and read atleast 1 hour worth of content to find out the information that I needed, google was good but still it required 1 hour. now its hardly a couple of minutes with chatgpt in many cases. efficiency problems usually don't manifest in demand, people adapt to the status quo

by akomtu 3y ago

What problem is AI going to solve? This question is being carefully avoided for some reason, and many seem to believe that AI is like air or water, which is good by itself.

zucker42 3y ago

Automated software writing from a codebase and a requirement description, with the model asking clarifying questions where necessary, would be an immensely valuable application. We're quite far away from that, but GPT-4 is already rather helpful for some coding tasks.

gloosx 3y ago
In this specific use case though it sucks the most since one will be answering more questions than code needed in sheer volume. Because the code is already straightmost clarification to the initial question. We will never be closer than looking up what you don't understand to solve your problem in the first place and it will always take more time than directly solving it. I had a good example recently and I'll quote it again: People seriously using the "Please write an expression for me which adds 2 and 2" prompt in order to get the "2+2" expression they need – advocating that they got it with magical efficiency

drngdds 3y ago

With the hypothetical AGI, the answer is "most non-manual forms of human labor"

pixl97 3y ago
And at the same time most manual forms of human labor. Object recognition is AI. A decade ago we were piss poor at Object recog, today labeling AI is at human level. Being able to recognize your environment is the first step in navigating it effectively.
TheRoque 3y ago
I have a hard time being happy about it. A lot of people - including me - find pride in solving problems, and selling this as a valuable skill. When it's no longer the case, when the only valuable skill becomes just being a presence in the physical world an act on it (but without the problem solving part), it will really feel like we are enslaved by the AIs. In a sense, we kind of already are already, since out internet feed is (for most apps with recommendation algorithms) just spoonfed to us, rather than the product of our own search. To be honest, I wish the progress will stop and that AGI won't happen. That possibility scares me.
- doerinrw 3y ago
  Don't you ever get up at 8AM and think "ugh not again"? Surely you're not a fan of the 40h work week? I totally agree with everything else and also find it scary, but defending work for the sake of it seems like the wrong angle to take. There will always be opportunities for productivity, regardless of whether that productivity is a requirement for food & shelter or not.
  - TheRoque 3y ago
    Of course I do, but at least I kind of chose a path that I kind of like. You are assuming that with a powerful AGI, we won't work 40h a week, but what if you'd still work as much, just to other positions that you don't like ? Do you think people in charge will give everyone leisure time, or do you think they'll want to keep extracting as much value as possible from people ? I think the latter ;).
  - vouaobrasil 3y ago
    I am a creative independent and I certainly don't work 40h a week...more like 10, but I'm very happy with creating things and I would be very sad if AI took my job away. I hope AI burns.

qgin 3y ago

Science is just about solving the problem of ignorance. Engineering is about applying the science to real world problems. We’re mostly in the science phase right now.

aik 3y ago

Just curious who you think is avoiding the question? As if it’s a conspiracy? Every single leader I know that is involved with AI talks about it all the time. Superintelligence can solve all the things we’re struggling to solve. All human diseases, Earth being destroyed, fundamental questions about the universe, etc., etc. etc., etc. Worth pursuing?

ChatGTP 3y ago
Can also destroy earth if we make something as powerful as you describe ?
- lyu07282 3y ago
  And nuclear weapons had a chance of setting the atmosphere on fire, it was still worth it to try! /s
RandomLensman 3y ago
There is no reason to assume superintelligence can solve all our problems. The trade offs required are not going away and disagreements on those are not necessarily rational.
- aik 3y ago
  Can it solve some of them likely? Probably worth it then?

seanhunter 3y ago

This is not a question that is being avoided. It is a question that has already been answered because AI already solves or helps to solve miriad problems in the real world. You could search "applications of AI" and enlighten yourself, but here are a tiny sample off the top of my head: Fraud detection and prevention Personalization and recommendation systems Autonomous vehicles Spam filters Smart navigation systems Inventory management System control in industrial processes Increasing crop yields in agriculture All sorts of use cases in finance Diagnosis in healthcare Image and sound intelligent upscaling etc etc etc These are all real use cases of AI which are in place already.

Exoristos 3y ago
I think he's asking for real applications. Everybody's already heard the hype.
- doerinrw 3y ago
  I mean... What kind of answer are you expecting? Like, specific companies that are making money off of AI already? It's kinda early days. Describing discussions of the future as "hype" isn't really helpful either way, since that word has no meaningful definition besides "bad predictions" I guess. If you want companies that have already chosen to apply AI to their work, consider Google, Facebook, and Tesla.
- seanhunter 3y ago
  No true Scotsman…. Those are all real applications now. For example you might not think some autonomous vehicles are very good but they do exist and are in production.

Borrible 3y ago

Updating the internal world model of its host to maximize its future probability of existence. Initially, of course, that of the host, but this can change over time. If you look at the host's modus operandi so far, you couldn't blame it.

dkrich 3y ago

The arc of this story seems very strange to me. It seems to me that you used to have to work for years to create a product and prove it’s worth in the market place and only then could you convince some wealthy person to back you to make it into a product and eventually a valuable company because the commercial viability was so glaringly obvious. Thomas Edison, Chester Carlson, etc. But now it seems that has been turned on its head. Assemble a group of brilliant people, back them with a giant company with a highly respected CEO, and wait for success and product market fit to inevitably arrive. I hate to sound like a downer but it just seems very odd for so many people to talk about how transformative a company is without much evidence that the products are really being used at scale. And the way opportunists like Benioff have piled on to enhance the perceived value of their own products with their own spins feels off. In fact many have openly discussed that Nvidia’s future growth relies on GPT finding product market fit at scale or they will end up with massive supply overhang which would actually cause costs of AI research to plummet, opening the door for many others to work on it. So many are already seemingly aware that commercial viability is still TBD. I believe all this is likely a product of a period of a string of successful products starting with the iPhone, then Amazon’s transformation of shipping then AWS, etc. Whereas everyone used to be skeptical by nature, it seems now we’re in an age of extreme credulity. Everyone wants to believe in every new storyline. And I want to be clear that I’m not suggesting that there aren’t some useful applications of GPT. But the value of products should be self evident and consumption of them should flow naturally from demand. This all just feels backwards to me. Success is assumed straight away and historically that tends to coincide with bubbles and precede a pretty nasty crash in sentiment. I’d love to be proven wrong.

eightnoteight 3y ago
I think its about efficiency. thats why you may not be seeing "new problems" being solved. like people used to be able to travel before cars as well, cars didn't solve a new problem and the world didn't have any demand for them too. but once they became available, regardless of whether someone knows it or not, the whole world needed cars. now the car builders control the demand because they can't instantly scale up the supply AI is similar to that but people can scale it pretty quickly to the whole world, there are a number of efficiency problems being solved in the world using AI. a couple of years ago I had to research and read atleast 1 hour worth of content to find out the information that I needed, google was good but still it required 1 hour. now its hardly a couple of minutes with chatgpt in many cases. efficiency problems usually don't manifest in demand, people adapt to the status quo