4 ms·

It is a real problem. The web will soon be full of AI generated bullshit. What might come back as useful tool? Authorship. People actually signing the content

by srge 4y ago

It is a real problem. The web will soon be full of AI generated bullshit. What might come back as useful tool? Authorship. People actually signing the content and putting their reputation on the line. It still exists obviously (journalism…) but it might need to be standardized and industrialized.

jasfi 4y ago
Another problem is AI training on AI generated content. This could lead to degradation of results. So perhaps to make everyone happy a new HTML tag indicating AI content would be useful?
- adql 4y ago
  The people using it to make money don't want it to be easily spotted, tag or metadata marking it will never happen unless required by law
  - jasfi 4y ago
    There would indeed need to be a law.
- Mordisquitos 4y ago
  > Another problem is AI training on AI generated content. This could lead to degradation of results. That's only the first step in the degradation. In a worse case scenario, where AI-generated crap is not contained, imagine this: young and still immature humans "learning" their styles of writing, information, and social expectations from content generated by AI that itself trained on AI-generated content─and then more AI trains on content generated by these humans that were trained on AI-generated content. And on, and on, and on...
  - unglaublich 4y ago
    Luckily, the phenomenon of puberty will make sure that young people absolutely do _not_ want to comply ;).
    - Mordisquitos 4y ago
      Ah, but they do want to fit in and speak like their peers, and while a significant part of their socialisation happens online, and as long as that is where they absorb most content, AI will soon infiltrate the circle of online strangers with whom they interact. That's all the more so considering that teenagers are prime targets for advertising networks of dubious ethical standards, who will be the first to deploy (or may already have deployed) well trained and self-correcting "How do you do, fellow kids?" NNs in potentially lucrative communities or social networks. (I'm not always as doom-and-gloom as it may seem, I'm just in a pessimistic mood today for some reason)
    - pixl97 4y ago
      Comply with what? Even in culture we tend to have a limited number of subsets of counter culture. The vast majority of the non-compliant to primary culture fits in one of a few common subcultures. As much as HN users try to exclaim they are islands that stand alone from society, we are all dependant on a functioning society for day to day life. We should ask questions like "Can countless AI's fracture society into a million sub societies disconnected from the reality around them".
- throwup 4y ago
  > a new HTML tag indicating AI content would be useful This has about as much chance of working as the evil bit. https://en.wikipedia.org/wiki/Evil_bit https://en.wikipedia.org/wiki/Evil_bit
  - jasfi 4y ago
    It worked for GDPR (regulation would be required).
- BeFlatXIII 4y ago
  That just makes things easier for the legitimate humans to differentiate themselves as being a cut above.
2devnull 4y ago
If you didn’t think the web wasn’t full of bullshit before you won’t be too disappointed.
- srge 4y ago
  Well I won’t be surprised if we all end up regretting the current level of bullshit. Fast.
  - 2devnull 4y ago
    Fair enough. It’s like the lump of labor fallacy, but for bullshit rather than labor. One could view that as growing pains, but it is still growth in some sense.
goodpoint 4y ago
> What might come back as useful tool? Authorship. That, and removing the financial incentives to attract traffic AKA ads.
- ghaff 4y ago
  Though the only way to do that is probably to end up in a world where most worthwhile content is locked behind hard paywalls so there's one class who pays a lot for gated content (which was more or less the historical state of affairs, public libraries notwithstanding) and the plebes have to filter through all the ad-laden crap.
rxhernandez 4y ago
People "sign" their content with their faces on TikTok; it doesn't stop the deluge of complete nonsense that's widely accepted on that platform (much to the frustration of many academics trying to combat it).
- srge 4y ago
  Arguably the non-sense videos on TikTok are not an existential threat on the web.
gshubert17 4y ago
Indeed, Neal Stephenson in _Anathem_ (2008), in describing an alternate world (in which his "reticulum" is our "network") wrote "Early in the Reticulum—thousands of years ago—it became almost useless because it was cluttered with faulty, obsolete, or downright misleading information." "So crap filtering became important. Businesses were built around it. ... " Generating crap "didn't really take off until the military got interested" in a program called "Artificial Inanity". The defenses that were developed back then now "work so well that, most of the time, the users of the Reticulum don't know it's there. Just as you are not aware of the millions of germs trying and failing to attack your body every moment of every day." A group of people (the "Ita") developed techniques for a parallel reticulum in which they could keep information they had determined to be reliable. When there was news on the reticulum, they might take a couple of days to do sanity-checking or fact-checking. I'm guessing there would need to be reputation monitoring and cryptographic signatures to maintain the integrity of their alternate web.
jillesvangurp 4y ago
It already has a lot of AI and garbage content. And in some cases it's state sponsored propaganda, companies pitching products, or marketing departments attempting some notion of virality. The point is that there are plenty of people already gaming the system by just flooding the internet with carefully designed noise to bend opinions this way or that way. "Fall, or dodge in heaven" by Neal Stephenson actually touches on this topic quite a bit and comes to a similar conclusion. Not his most accessible work but I enjoyed it. That book was written just before Trump won 2016 elections and in hind sight making that book really spot on and timely. In the book, different streams of truth become so hard to separate due to all the spam, automated content, etc. that people start filtering it based on reputation and signatures. The game then becomes curating and filtering the content, which is ultimately not an objective thing but a subjective thing. So the net result is what we are actually starting to see today which is different groups of people locked in their own bubbles and no longer even getting exposed to the other side's point of view or even able to agree on some basic facts. Like who actually won the election. And people are feeding those bubbles with nonsense to influence opinions, get people to vote certain ways, or simply to just create chaos and fear. In the book the topic is not an election but the seemingly simple question of whether or not a certain town in the us still exists after it was supposedly blown up with a nuclear bomb. In the book, this event leads to people acting as if that happened and ignoring all evidence to the contrary pointing out that in fact this was an elaborate hoax designed to create a massive panic. That particular dystopian outcome is not necessarily what would happen. But using technology to authenticate where content actually comes from along with some reputation mechanism would go a long way separating the noise from the genuine content. It's simple and easy to do. So, why don't we?

by srge 4y ago

jasfi 4y ago

Another problem is AI training on AI generated content. This could lead to degradation of results. So perhaps to make everyone happy a new HTML tag indicating AI content would be useful?

adql 4y ago
The people using it to make money don't want it to be easily spotted, tag or metadata marking it will never happen unless required by law
- jasfi 4y ago
  There would indeed need to be a law.
Mordisquitos 4y ago
> Another problem is AI training on AI generated content. This could lead to degradation of results. That's only the first step in the degradation. In a worse case scenario, where AI-generated crap is not contained, imagine this: young and still immature humans "learning" their styles of writing, information, and social expectations from content generated by AI that itself trained on AI-generated content─and then more AI trains on content generated by these humans that were trained on AI-generated content. And on, and on, and on...
- unglaublich 4y ago
  Luckily, the phenomenon of puberty will make sure that young people absolutely do _not_ want to comply ;).
  - Mordisquitos 4y ago
    Ah, but they do want to fit in and speak like their peers, and while a significant part of their socialisation happens online, and as long as that is where they absorb most content, AI will soon infiltrate the circle of online strangers with whom they interact. That's all the more so considering that teenagers are prime targets for advertising networks of dubious ethical standards, who will be the first to deploy (or may already have deployed) well trained and self-correcting "How do you do, fellow kids?" NNs in potentially lucrative communities or social networks. (I'm not always as doom-and-gloom as it may seem, I'm just in a pessimistic mood today for some reason)
  - pixl97 4y ago
    Comply with what? Even in culture we tend to have a limited number of subsets of counter culture. The vast majority of the non-compliant to primary culture fits in one of a few common subcultures. As much as HN users try to exclaim they are islands that stand alone from society, we are all dependant on a functioning society for day to day life. We should ask questions like "Can countless AI's fracture society into a million sub societies disconnected from the reality around them".
throwup 4y ago
> a new HTML tag indicating AI content would be useful This has about as much chance of working as the evil bit. https://en.wikipedia.org/wiki/Evil_bit https://en.wikipedia.org/wiki/Evil_bit
- jasfi 4y ago
  It worked for GDPR (regulation would be required).
BeFlatXIII 4y ago
That just makes things easier for the legitimate humans to differentiate themselves as being a cut above.

2devnull 4y ago

If you didn’t think the web wasn’t full of bullshit before you won’t be too disappointed.

srge 4y ago
Well I won’t be surprised if we all end up regretting the current level of bullshit. Fast.
- 2devnull 4y ago
  Fair enough. It’s like the lump of labor fallacy, but for bullshit rather than labor. One could view that as growing pains, but it is still growth in some sense.

goodpoint 4y ago

> What might come back as useful tool? Authorship. That, and removing the financial incentives to attract traffic AKA ads.

ghaff 4y ago
Though the only way to do that is probably to end up in a world where most worthwhile content is locked behind hard paywalls so there's one class who pays a lot for gated content (which was more or less the historical state of affairs, public libraries notwithstanding) and the plebes have to filter through all the ad-laden crap.

rxhernandez 4y ago

People "sign" their content with their faces on TikTok; it doesn't stop the deluge of complete nonsense that's widely accepted on that platform (much to the frustration of many academics trying to combat it).

srge 4y ago
Arguably the non-sense videos on TikTok are not an existential threat on the web.

gshubert17 4y ago

Indeed, Neal Stephenson in _Anathem_ (2008), in describing an alternate world (in which his "reticulum" is our "network") wrote "Early in the Reticulum—thousands of years ago—it became almost useless because it was cluttered with faulty, obsolete, or downright misleading information." "So crap filtering became important. Businesses were built around it. ... " Generating crap "didn't really take off until the military got interested" in a program called "Artificial Inanity". The defenses that were developed back then now "work so well that, most of the time, the users of the Reticulum don't know it's there. Just as you are not aware of the millions of germs trying and failing to attack your body every moment of every day." A group of people (the "Ita") developed techniques for a parallel reticulum in which they could keep information they had determined to be reliable. When there was news on the reticulum, they might take a couple of days to do sanity-checking or fact-checking. I'm guessing there would need to be reputation monitoring and cryptographic signatures to maintain the integrity of their alternate web.

jillesvangurp 4y ago

It already has a lot of AI and garbage content. And in some cases it's state sponsored propaganda, companies pitching products, or marketing departments attempting some notion of virality. The point is that there are plenty of people already gaming the system by just flooding the internet with carefully designed noise to bend opinions this way or that way. "Fall, or dodge in heaven" by Neal Stephenson actually touches on this topic quite a bit and comes to a similar conclusion. Not his most accessible work but I enjoyed it. That book was written just before Trump won 2016 elections and in hind sight making that book really spot on and timely. In the book, different streams of truth become so hard to separate due to all the spam, automated content, etc. that people start filtering it based on reputation and signatures. The game then becomes curating and filtering the content, which is ultimately not an objective thing but a subjective thing. So the net result is what we are actually starting to see today which is different groups of people locked in their own bubbles and no longer even getting exposed to the other side's point of view or even able to agree on some basic facts. Like who actually won the election. And people are feeding those bubbles with nonsense to influence opinions, get people to vote certain ways, or simply to just create chaos and fear. In the book the topic is not an election but the seemingly simple question of whether or not a certain town in the us still exists after it was supposedly blown up with a nuclear bomb. In the book, this event leads to people acting as if that happened and ignoring all evidence to the contrary pointing out that in fact this was an elaborate hoax designed to create a massive panic. That particular dystopian outcome is not necessarily what would happen. But using technology to authenticate where content actually comes from along with some reputation mechanism would go a long way separating the noise from the genuine content. It's simple and easy to do. So, why don't we?