6 ms·

I agree. However the ChatGPT situation is the first hype that’s really been harmful rather than just disruptive. It can and is being used to generate low accura

by unxdfa 4y ago

I agree. However the ChatGPT situation is the first hype that’s really been harmful rather than just disruptive. It can and is being used to generate low accuracy noise which degrades quality information. I suspect our demise isn’t going to be a big bang but a slow fizzle of information decay. And this is the start of it. At best it’s going to make us stupider.

fock 4y ago
recently I tried to search for CO2-consumption (as in usage as an industrial gas) on Google. It didn't return 1 useful hit, everything was about emissions. So: We are already there, all without ChatGPT... ChatGPT (with its cool ways to make it nondeterministic, like: replacing tokens with synonyms and raising the temperature. Or: retraining it...) will make things worse definitely. Probably the way to go is for {academia,sects,orgs with sense for humanity,...} to build stores of attributable information and then start shunning the branches which produce bullshit (this is what scientific publishing should do already.... it doesn't... even without ChatGPT there's already templated articles strewn around the corpus and noone cares). The main issue there is, that this means that they will also have to stop using a lot of technology (and buy a 32nm foundry...), because many orgs are probably already poisoned.
- mtlmtlmtlmtl 4y ago
  I went and searched for "CO2 industrial applications -emissions" without the quotes. This was the first result: https://applications.messergroup.com/co2-applications https://applications.messergroup.com/co2-applications There were a number of other relevant results. Papers on various applications of CO2, forum posts, etc. Nothing on emissions. Sure google is a lot worse these days than it used to be, but it still seems plenty usable to me and I have no issues finding what I want with some basic operators.
  - Crosseye_Jack 4y ago
    > with some basic operators The problem imo is that Google was good enough that basic operators were not required, as such the avg Joe is amazed when finding out that such operators exist, and Google doesn’t really promote the advanced search page these days.
    - yunohn 4y ago
      I’m not sure why GP used the minus operator - I omitted that part and still got results about usage and not emissions.
      - fock 4y ago
        I want to get statistics and numbers. I'm aware of the applications. GPs suggestions didn't really help me either.
        mtlmtlmtlmtl 4y ago
        Did another search for "industrial consumption of CO2", again without the quotes, and found this: https://www.iea.org/reports/putting-co2-to-use https://www.iea.org/reports/putting-co2-to-use Sure this talks a lot about emissions, but it also seems to include a detailed breakdown of various industrial uses for CO2, with numbers, projections, statistics and all.
        yunohn 4y ago
        https://www.iea.org/reports/putting-co2-to-use https://www.iea.org/reports/putting-co2-to-use First result for “co2 industrial usage numbers”, without quotes - just like the sibling comment. You need to specify atleast a few keywords to clarify what you’re looking for - search engine nor llm chatbot nor HN commenters can read your mind…
      - mtlmtlmtlmtl 4y ago
        Specifically because they said they only found stuff on emissions. Never tried it without. It's one of the most common patterns I come across. I search word X but find a bunch of word Y stuff that's not what I'm looking for, because word X and Y appear together a lot in Y's context. So I add -Y to the query.
    - msrenee 4y ago
      Google has gotten way better at interpreting search queries formulated like natural speech. It used to be you had to know what sort of keywords together would get you the results you were interested in. Now I can type in "what is the capital of France" and have a Wikipedia article about Paris pop up, an Encyclopedia Brittanica entry, and World Population Review for the city. That wouldn't have been the case in 2005. It's become a tool that basically anyone can at least get some information out of. Googling used to be a skill that non-technical people struggled with.
raspberry1337 4y ago
The solution is already predicted by scifi. We will start building new sub-internets with stricter, formalised, rules to reduce spam and noise. The outside internet will become the wild, noisy internet where few people reside. It's like in Cyberpunk 2077 where there is the "wall" that keeps all the AI out from the mainstream internet.
- jonathanstrange 4y ago
  I've been thinking for a while about writing another network on top of IP with decentralized DNS (dealing with domain names similarly to how SSH deals with public keys), make it open-source, free, with a one-click installer, and strictly prohibit any commercial activity on it (no ads, no acting on behalf of a company) or at least make commercial activities selectively opt-in for the end user. The same could be done for bots. Some of the technical aspect are already solved by others but the problem is that such a project would need some larger organizational support with lawyers in order to make the EULA enforceable. In other words, it's more of a social than a technical issue.
  - hnlmorg 4y ago
    I don’t think an EULA will stop bots, spam and scams. And without commercial buy in, I can’t see why anyone apart from computer geeks and scammers would want to use it. The first question your average users would ask is “why isn’t $SHOP on there”. Much as I dislike the modern web, it’s what the people have chosen in exchange for free services (like email) and convenience (like online shopping).
    - jonathanstrange 4y ago
      I don't care at all about average users, this network would be an alternative on top of what we have and can be as elitist as the small user group wants. The difficult part is having the power to sue organizations who violate the EULA (a social and financial problem) and getting a decentralized banning system to work (a very tricky technical, distributed social choice problem).
blibble 4y ago
> However the ChatGPT situation is the first hype that’s really been harmful rather than just disruptive. blockchain? all that pointless CO2 for nothing
- unxdfa 4y ago
  That's true. Agree there.
chiefalchemist 4y ago
Noise? We've been suffering from it quite some time now. Once complete control (of information) wasn't possible the next choice has been baffle them with bullshit (and repetition). This is happening on all fronts, all ideologies. Sure some may be more subtle than others but it's everywhere. To believe otherwise is simply naive. ChatGPT is a symptom that perhaps finally raises discussion of the root problem above the noise. Yeah, ironic.
lakomen 4y ago
Information decay has started around the year 2000. I'm not a Hipster, I have experienced the internet pre 2000, which was an semi exclusive club for educated and intelligent people with accurate search results. After the masses poured onto it the results became more half truths and opinions. And now you have "AI" feeding on those half truths and opinions. Whatever could go wrong? However imagine if it was fed with facts only.
- jvanderbot 4y ago
  Well, you can't train LLM on "facts alone", since they are basically throwing darts at the next word, not using information to craft a message. But I hear your larger point: it'd be nice to have an agent that was trained on wikipedia information, but already spoke a language
  - r3trohack3r 4y ago
    GPT is trained on Wikipedia though. It’s speculated libgen is in there as well.
    - artificial 4y ago
      Let’s not forget Twitter was a training source. Early Twitter was like 4chan.
    - jvanderbot 4y ago
      I think GGP comment started this thread about a knowledge engine, not a "word generator" if that makes sense. Training on Wikipedia in this case means "training to put together words like wikipedia authors do" right? Not "learning the facts of wikipedia to provide useful synthesis and cross referencing." though it probably does OK by accident.
- ethbr0 4y ago
  Imho, info decay took off when social metastasized, which is to say ~2010+. It was an interesting time, because (from memory) 2000 was still "It's pretty complicated for average person to create and post content on the web" and transitioned to 2010's "There are highly polished, easy to use, free SaaS solutions that a trained dog could use." With mobile versions of same in the next few years. (All dates mass market penetration, not first release) There was a direct correlation between ease-of-publishing and less intellectual discourse on the web. "Info decay" was definitely a good way to put it. It wasn't that people were necessarily dumb or angry, but that there was a lot more honestly incorrect information online. Before, there were crackpots, but it was pretty easy to tell they were crazy. Having seen both, the banal wrongness is far worse. I think that's one reason wikipedia really took off ~2005+. People were craving an easier way to get even roughly-right information.
Dalewyn 4y ago
> It can and is being used to generate low accuracy noise which degrades quality information. I would argue this is a problem stemming from the way "AI" is presented rather than anything to do with "AI" itself. That is to say, fucking nobody trusts autocorrect to be correct and "AI" is really just a more complex form of autocorrect. But people hear "AI" and think AI as in Transformers, Terminator, R2-D2, etc. They're intelligent, right? And that's where the presentation is doing significant harm. There is nothing intelligent, let alone sentient, about "AI".
lynx23 4y ago
Information decay has started with forums for the masses. It has become increasingly hard to actually find useful advice, because lots of the "answers" you find are actually partly wrong, mostly because people without knowhow feel like they need to answer questions to boost their karma... Many of my coworkers feel the same. And this started way before GPT was conceived. Besides, your post somehow feels like generic technology-angst to me. It seems like whenever we invent something new, there are those people that predict this is finally our demise... GGPT is a tool, like every other technology. Use it wisely, and you will likely benefit. Use it without engaging your brain, and you will like see negative effects.

by unxdfa 4y ago

fock 4y ago

recently I tried to search for CO2-consumption (as in usage as an industrial gas) on Google. It didn't return 1 useful hit, everything was about emissions. So: We are already there, all without ChatGPT... ChatGPT (with its cool ways to make it nondeterministic, like: replacing tokens with synonyms and raising the temperature. Or: retraining it...) will make things worse definitely. Probably the way to go is for {academia,sects,orgs with sense for humanity,...} to build stores of attributable information and then start shunning the branches which produce bullshit (this is what scientific publishing should do already.... it doesn't... even without ChatGPT there's already templated articles strewn around the corpus and noone cares). The main issue there is, that this means that they will also have to stop using a lot of technology (and buy a 32nm foundry...), because many orgs are probably already poisoned.

mtlmtlmtlmtl 4y ago
I went and searched for "CO2 industrial applications -emissions" without the quotes. This was the first result: https://applications.messergroup.com/co2-applications https://applications.messergroup.com/co2-applications There were a number of other relevant results. Papers on various applications of CO2, forum posts, etc. Nothing on emissions. Sure google is a lot worse these days than it used to be, but it still seems plenty usable to me and I have no issues finding what I want with some basic operators.
- Crosseye_Jack 4y ago
  > with some basic operators The problem imo is that Google was good enough that basic operators were not required, as such the avg Joe is amazed when finding out that such operators exist, and Google doesn’t really promote the advanced search page these days.
  - yunohn 4y ago
    I’m not sure why GP used the minus operator - I omitted that part and still got results about usage and not emissions.
    - fock 4y ago
      I want to get statistics and numbers. I'm aware of the applications. GPs suggestions didn't really help me either.
      - mtlmtlmtlmtl 4y ago
        Did another search for "industrial consumption of CO2", again without the quotes, and found this: https://www.iea.org/reports/putting-co2-to-use https://www.iea.org/reports/putting-co2-to-use Sure this talks a lot about emissions, but it also seems to include a detailed breakdown of various industrial uses for CO2, with numbers, projections, statistics and all.
      - yunohn 4y ago
        https://www.iea.org/reports/putting-co2-to-use https://www.iea.org/reports/putting-co2-to-use First result for “co2 industrial usage numbers”, without quotes - just like the sibling comment. You need to specify atleast a few keywords to clarify what you’re looking for - search engine nor llm chatbot nor HN commenters can read your mind…
    - mtlmtlmtlmtl 4y ago
      Specifically because they said they only found stuff on emissions. Never tried it without. It's one of the most common patterns I come across. I search word X but find a bunch of word Y stuff that's not what I'm looking for, because word X and Y appear together a lot in Y's context. So I add -Y to the query.
  - msrenee 4y ago
    Google has gotten way better at interpreting search queries formulated like natural speech. It used to be you had to know what sort of keywords together would get you the results you were interested in. Now I can type in "what is the capital of France" and have a Wikipedia article about Paris pop up, an Encyclopedia Brittanica entry, and World Population Review for the city. That wouldn't have been the case in 2005. It's become a tool that basically anyone can at least get some information out of. Googling used to be a skill that non-technical people struggled with.

raspberry1337 4y ago

The solution is already predicted by scifi. We will start building new sub-internets with stricter, formalised, rules to reduce spam and noise. The outside internet will become the wild, noisy internet where few people reside. It's like in Cyberpunk 2077 where there is the "wall" that keeps all the AI out from the mainstream internet.

jonathanstrange 4y ago
I've been thinking for a while about writing another network on top of IP with decentralized DNS (dealing with domain names similarly to how SSH deals with public keys), make it open-source, free, with a one-click installer, and strictly prohibit any commercial activity on it (no ads, no acting on behalf of a company) or at least make commercial activities selectively opt-in for the end user. The same could be done for bots. Some of the technical aspect are already solved by others but the problem is that such a project would need some larger organizational support with lawyers in order to make the EULA enforceable. In other words, it's more of a social than a technical issue.
- hnlmorg 4y ago
  I don’t think an EULA will stop bots, spam and scams. And without commercial buy in, I can’t see why anyone apart from computer geeks and scammers would want to use it. The first question your average users would ask is “why isn’t $SHOP on there”. Much as I dislike the modern web, it’s what the people have chosen in exchange for free services (like email) and convenience (like online shopping).
  - jonathanstrange 4y ago
    I don't care at all about average users, this network would be an alternative on top of what we have and can be as elitist as the small user group wants. The difficult part is having the power to sue organizations who violate the EULA (a social and financial problem) and getting a decentralized banning system to work (a very tricky technical, distributed social choice problem).

blibble 4y ago

> However the ChatGPT situation is the first hype that’s really been harmful rather than just disruptive. blockchain? all that pointless CO2 for nothing

unxdfa 4y ago
That's true. Agree there.

chiefalchemist 4y ago

Noise? We've been suffering from it quite some time now. Once complete control (of information) wasn't possible the next choice has been baffle them with bullshit (and repetition). This is happening on all fronts, all ideologies. Sure some may be more subtle than others but it's everywhere. To believe otherwise is simply naive. ChatGPT is a symptom that perhaps finally raises discussion of the root problem above the noise. Yeah, ironic.

lakomen 4y ago

Information decay has started around the year 2000. I'm not a Hipster, I have experienced the internet pre 2000, which was an semi exclusive club for educated and intelligent people with accurate search results. After the masses poured onto it the results became more half truths and opinions. And now you have "AI" feeding on those half truths and opinions. Whatever could go wrong? However imagine if it was fed with facts only.

jvanderbot 4y ago
Well, you can't train LLM on "facts alone", since they are basically throwing darts at the next word, not using information to craft a message. But I hear your larger point: it'd be nice to have an agent that was trained on wikipedia information, but already spoke a language
- r3trohack3r 4y ago
  GPT is trained on Wikipedia though. It’s speculated libgen is in there as well.
  - artificial 4y ago
    Let’s not forget Twitter was a training source. Early Twitter was like 4chan.
  - jvanderbot 4y ago
    I think GGP comment started this thread about a knowledge engine, not a "word generator" if that makes sense. Training on Wikipedia in this case means "training to put together words like wikipedia authors do" right? Not "learning the facts of wikipedia to provide useful synthesis and cross referencing." though it probably does OK by accident.
ethbr0 4y ago
Imho, info decay took off when social metastasized, which is to say ~2010+. It was an interesting time, because (from memory) 2000 was still "It's pretty complicated for average person to create and post content on the web" and transitioned to 2010's "There are highly polished, easy to use, free SaaS solutions that a trained dog could use." With mobile versions of same in the next few years. (All dates mass market penetration, not first release) There was a direct correlation between ease-of-publishing and less intellectual discourse on the web. "Info decay" was definitely a good way to put it. It wasn't that people were necessarily dumb or angry, but that there was a lot more honestly incorrect information online. Before, there were crackpots, but it was pretty easy to tell they were crazy. Having seen both, the banal wrongness is far worse. I think that's one reason wikipedia really took off ~2005+. People were craving an easier way to get even roughly-right information.

Dalewyn 4y ago

> It can and is being used to generate low accuracy noise which degrades quality information. I would argue this is a problem stemming from the way "AI" is presented rather than anything to do with "AI" itself. That is to say, fucking nobody trusts autocorrect to be correct and "AI" is really just a more complex form of autocorrect. But people hear "AI" and think AI as in Transformers, Terminator, R2-D2, etc. They're intelligent, right? And that's where the presentation is doing significant harm. There is nothing intelligent, let alone sentient, about "AI".

lynx23 4y ago

Information decay has started with forums for the masses. It has become increasingly hard to actually find useful advice, because lots of the "answers" you find are actually partly wrong, mostly because people without knowhow feel like they need to answer questions to boost their karma... Many of my coworkers feel the same. And this started way before GPT was conceived. Besides, your post somehow feels like generic technology-angst to me. It seems like whenever we invent something new, there are those people that predict this is finally our demise... GGPT is a tool, like every other technology. Use it wisely, and you will likely benefit. Use it without engaging your brain, and you will like see negative effects.