4 ms·

Yeah it seems likely to me to be an overzealous AI decision, given the background noise of anti-vaccine disinformation on youtube these days I wouldn't be surpr

by mojzu 5y ago

Yeah it seems likely to me to be an overzealous AI decision, given the background noise of anti-vaccine disinformation on youtube these days I wouldn't be surprised if they've turned up their algorithms at the cost of more false and marginal positives What's strange however is them seeming to use this to try and 'strengthen' the online safety bill which in it's current form would already be damaging to free speech online (adds massive liability for illegal/'harmful' content if it isn't removed quickly so companies will most likely block more content as a result)

PontifexMinimus 5y ago
> an overzealous AI decision Then they simply shouldn't use AI algorithms to censor people. > What's strange however is them seeming to use this to try and 'strengthen' the online safety bill which in it's current form would already be damaging to free speech online It's not obvious to me that the UK government has any coherent idea of what it is trying to do.
- ColinHayhurst 5y ago
  Some would say the Bill is a symptom of Facebook Derangement Syndrome https://webdevlaw.uk/2020/12/09/facebook-derangement-syndrome/ https://webdevlaw.uk/2020/12/09/facebook-derangement-syndrom... Based on those they are consulting with that seems justfied.
mathw 5y ago
Sadly that is not strange - if there are any technologically literate people advising our government on policy they're not being listened to.
hliyan 5y ago
Should we be more forgiving of harm done by faulty algorithms than by negligent human beings? It was the company's choice to save money by automating a process that normally only human beings can do reliably. The consequences to the victim are the same regardless of the instrument used in the commission of the wrongdoing.
- mojzu 5y ago
  Given the quantity of video uploaded to youtube I'm not sure human moderation of any significant amount of it is even possible anymore. Humans are definitely better at handling many situations but are also prone to their own biases (e.g. there were facebook moderators a while back who would take down reported content they disagreed with for religious reasons), they are often not given enough time to reasonably review a report, and when exposed to the kind of videos being reported can experience serious mental health problems Some of these could be dealt with by the companies investing more in moderation (make them employees/better pay/healthcare etc.), but if you want to moderate at the scale of youtube/facebook/etc. then I don't see how you could do it without some automated tools
  - ClumsyPilot 5y ago
    "human moderation of any significant amount of it is even possible anymore." Then maybe free video hosting is not a viable business, why should youtube destroy all competition and then complain that checking their uploads is not possoble?
    - hliyan 5y ago
      Exactly. Perhaps if lots of little video hosting companies existed, this wouldn't be an issue. "Too big to do the right thing" seems like an even worse version of "Too big to fail".
    - mojzu 5y ago
      The argument that these companies are too big and that creates bad incentives/lack of incentives to improve is one I'm much more sympathetic to. I'd be in favour of more smaller companies/communities although I don't think this would remove the need for automated tools (e.g. smallish reddit and discord communities often use bots and other tools to automate some moderation). And to make something like this a reality you'd probably need government intervention to overcome the network effect that these bigger platforms have
  - hliyan 5y ago
    My personal belief is "that won't scale" is not a good reason for bypassing ethical or socially responsible behavior.

by mojzu 5y ago

PontifexMinimus 5y ago

> an overzealous AI decision Then they simply shouldn't use AI algorithms to censor people. > What's strange however is them seeming to use this to try and 'strengthen' the online safety bill which in it's current form would already be damaging to free speech online It's not obvious to me that the UK government has any coherent idea of what it is trying to do.

ColinHayhurst 5y ago
Some would say the Bill is a symptom of Facebook Derangement Syndrome https://webdevlaw.uk/2020/12/09/facebook-derangement-syndrome/ https://webdevlaw.uk/2020/12/09/facebook-derangement-syndrom... Based on those they are consulting with that seems justfied.

mathw 5y ago

Sadly that is not strange - if there are any technologically literate people advising our government on policy they're not being listened to.

hliyan 5y ago

Should we be more forgiving of harm done by faulty algorithms than by negligent human beings? It was the company's choice to save money by automating a process that normally only human beings can do reliably. The consequences to the victim are the same regardless of the instrument used in the commission of the wrongdoing.

mojzu 5y ago
Given the quantity of video uploaded to youtube I'm not sure human moderation of any significant amount of it is even possible anymore. Humans are definitely better at handling many situations but are also prone to their own biases (e.g. there were facebook moderators a while back who would take down reported content they disagreed with for religious reasons), they are often not given enough time to reasonably review a report, and when exposed to the kind of videos being reported can experience serious mental health problems Some of these could be dealt with by the companies investing more in moderation (make them employees/better pay/healthcare etc.), but if you want to moderate at the scale of youtube/facebook/etc. then I don't see how you could do it without some automated tools
- ClumsyPilot 5y ago
  "human moderation of any significant amount of it is even possible anymore." Then maybe free video hosting is not a viable business, why should youtube destroy all competition and then complain that checking their uploads is not possoble?
  - hliyan 5y ago
    Exactly. Perhaps if lots of little video hosting companies existed, this wouldn't be an issue. "Too big to do the right thing" seems like an even worse version of "Too big to fail".
  - mojzu 5y ago
    The argument that these companies are too big and that creates bad incentives/lack of incentives to improve is one I'm much more sympathetic to. I'd be in favour of more smaller companies/communities although I don't think this would remove the need for automated tools (e.g. smallish reddit and discord communities often use bots and other tools to automate some moderation). And to make something like this a reality you'd probably need government intervention to overcome the network effect that these bigger platforms have
- hliyan 5y ago
  My personal belief is "that won't scale" is not a good reason for bypassing ethical or socially responsible behavior.