4 ms·

> Not sure where the next revolution may come from (no, 'AI' is not it). Not sure why you say this; are you sure AI can or should be so easily dismissed out of

by unoti 3y ago

> Not sure where the next revolution may come from (no, 'AI' is not it). Not sure why you say this; are you sure AI can or should be so easily dismissed out of hand without further thoughts? --- The assertion that "AI is not it" when considering the next revolution is a point of debate. While it's true that AI hasn't yet led to a socio-economic transformation on the scale of the internet or mobile technology, it is still in the early stages of its potential impact. AI and machine learning are already influencing many industries, from healthcare with predictive diagnostics to transportation with autonomous vehicles, and their capabilities are rapidly advancing. The part below "---" above was written by ChatGPT, to help underscore my point.

emodendroket 3y ago
Yeah, the part below the dashes does seem pretty derivative now that you mention it.
somewhereoutth 3y ago
Most likely LLMs (and this is what we mean when we talk about the latest AI 'craze') will find useful purposes in some situations, almost always as an adjunct to a human (see Centaur Chess). Perhaps LLMs' biggest contribution to social progress will be wrecking social media, as you have kindly demonstrated. Nobody wants to talk to a machine.
- ianbutler 3y ago
  > "Nobody wants to talk to a machine." "ChatGPT has 180.5 million users, generates $80 million/month revenue, and crossed over 10 billion all-time visits (1.43B visits in August 2023)" If you can't understand why normal people like having a human like chat interface to a machine then you're a bit out of touch.
  - somewhereoutth 3y ago
    I don't actually know any normal (i.e. outside IT) people that have talked to or are even slightly interested in talking to ChatGPT. The user figures will have been inflated as they are with any public facing platform, the money is largely VCs jumping on the hype, and the all-time visits will be pumped by automated services attempting (and likely failing) to demonstrate value.
    - ianbutler 3y ago
      Weird, I do know normal people who do. I also know businesses who have it integrated with normal people talking to it. Personally, I think you're just cynical here. I'm not going to choose to take your opinion on this just because when I continue to see concrete evidence to the contrary.
    - chezelenkoooo 3y ago
      I mean, even if you take one tenth of those figures that's still 18 million people.
  - outside1234 3y ago
    Now do profit
deathanatos 3y ago
… and GPT's rebuttal is basically content-free. AI has been in "the early stages" for a while now. ELIZA is more than half a century ago, and AI continues to make the basic inability-to-maintain-context level errors now that it did then. E.g., in a recent conversation with an LLM, I was consistently getting switched to the other gender by the LLM, even after attempting to correct it. In adventure stories, it'll make trivial context mistakes such as having characters whose hands are bound or otherwise incapable of motion do things that require those, e.g., pick up swords. Characters will swap or pop in or out of existence. In another example, I asked an AI recently for Kubernetes debugging advice, and after informing it that its suggestions didn't work, and asking "what would you try next?" … it repeated the same wrong suggestions. (I am amazed that "what is the luminosity of the sun?" is finally answered not with "1 solar luminosity", which it was for a while. Now, if we can get "what is my ipv4 address" to 1-box, Goog.) The sensor under the sink can't reliably suss out "hands" vs. "not hands", though I grant its probably a fair bit simpler in its construction … but that's sort of the point, too. --- What I would call basic, simple problems in health tech that we don't need AI to fix: … my own healthcare provider doesn't know who my GP with them is. My insurance doesn't know if the bill has been paid or not. My provider can't reliably determine that I am insured, or by who. Providers & insurers can't agree on what the bill is, or what is being billed, and are not capable of figuring it out. My insurer doesn't know what providers are in their network. We still transfer health information over god-awful protocols like HL7v2 that just destroy my time as a dev supporting them. … but yes, let's just keep trying to jam AI in somewhere, anywhere. Automotive is the same: see the self-driving car that crushed a woman recently resulting in the suspension of the parent company's license. Car prices are through the roof, to the point where it's not economically practical to purchase at the moment, and so the end result is … I'm still driving, by hand. They're advancing, I'll give them that, but I could feel far more meaningful impact in many places in my life from just the most basic of software engineering … or even just the most basic procedural changes (many of the problems above aren't even tech problems, but administrative ones).

by unoti 3y ago

emodendroket 3y ago

Yeah, the part below the dashes does seem pretty derivative now that you mention it.

somewhereoutth 3y ago

Most likely LLMs (and this is what we mean when we talk about the latest AI 'craze') will find useful purposes in some situations, almost always as an adjunct to a human (see Centaur Chess). Perhaps LLMs' biggest contribution to social progress will be wrecking social media, as you have kindly demonstrated. Nobody wants to talk to a machine.

ianbutler 3y ago
> "Nobody wants to talk to a machine." "ChatGPT has 180.5 million users, generates $80 million/month revenue, and crossed over 10 billion all-time visits (1.43B visits in August 2023)" If you can't understand why normal people like having a human like chat interface to a machine then you're a bit out of touch.
- somewhereoutth 3y ago
  I don't actually know any normal (i.e. outside IT) people that have talked to or are even slightly interested in talking to ChatGPT. The user figures will have been inflated as they are with any public facing platform, the money is largely VCs jumping on the hype, and the all-time visits will be pumped by automated services attempting (and likely failing) to demonstrate value.
  - ianbutler 3y ago
    Weird, I do know normal people who do. I also know businesses who have it integrated with normal people talking to it. Personally, I think you're just cynical here. I'm not going to choose to take your opinion on this just because when I continue to see concrete evidence to the contrary.
  - chezelenkoooo 3y ago
    I mean, even if you take one tenth of those figures that's still 18 million people.
- outside1234 3y ago
  Now do profit

deathanatos 3y ago

… and GPT's rebuttal is basically content-free. AI has been in "the early stages" for a while now. ELIZA is more than half a century ago, and AI continues to make the basic inability-to-maintain-context level errors now that it did then. E.g., in a recent conversation with an LLM, I was consistently getting switched to the other gender by the LLM, even after attempting to correct it. In adventure stories, it'll make trivial context mistakes such as having characters whose hands are bound or otherwise incapable of motion do things that require those, e.g., pick up swords. Characters will swap or pop in or out of existence. In another example, I asked an AI recently for Kubernetes debugging advice, and after informing it that its suggestions didn't work, and asking "what would you try next?" … it repeated the same wrong suggestions. (I am amazed that "what is the luminosity of the sun?" is finally answered not with "1 solar luminosity", which it was for a while. Now, if we can get "what is my ipv4 address" to 1-box, Goog.) The sensor under the sink can't reliably suss out "hands" vs. "not hands", though I grant its probably a fair bit simpler in its construction … but that's sort of the point, too. --- What I would call basic, simple problems in health tech that we don't need AI to fix: … my own healthcare provider doesn't know who my GP with them is. My insurance doesn't know if the bill has been paid or not. My provider can't reliably determine that I am insured, or by who. Providers & insurers can't agree on what the bill is, or what is being billed, and are not capable of figuring it out. My insurer doesn't know what providers are in their network. We still transfer health information over god-awful protocols like HL7v2 that just destroy my time as a dev supporting them. … but yes, let's just keep trying to jam AI in somewhere, anywhere. Automotive is the same: see the self-driving car that crushed a woman recently resulting in the suspension of the parent company's license. Car prices are through the roof, to the point where it's not economically practical to purchase at the moment, and so the end result is … I'm still driving, by hand. They're advancing, I'll give them that, but I could feel far more meaningful impact in many places in my life from just the most basic of software engineering … or even just the most basic procedural changes (many of the problems above aren't even tech problems, but administrative ones).