3 ms·

A real shocker for me is how little the big tech companies are using their AI powers for defensive purposes. Each smart-project they undertake to detect your fr

A real shocker for me is how little the big tech companies are using their AI powers for defensive purposes. Each smart-project they undertake to detect your friends in your photos and identify cats in pictures is done for a directly profitable feature. But with Twitter and Facebook both getting caught letting foreign actors buy political ads and use fake personas to influence the population, it seems like an existential risk to them. Wouldn't it be wise to direct some of this work to detecting fake accounts, providing an honest experience to users and being more honest overall? Yeah, Twitter and Facebook would show slower growth numbers. But they would become respectable and sustainable for society, which seems massively more valuable financially in the long run.

jsnell 9y ago
Just how do you know what these companies are using for fighting abuse (fake accounts, spam, account hijacking, etc)? Dealing with abuse is an adversarial process, where both sides are trying to adapt to what the other is doing. It's not like most programming where you can come up with a provably good enough solution. It also means that time is a valuable resource. There's nothing to be gained by e.g. Twitter announcing that they're now using ML on the profile images to detect fake accounts. That'd let the bad guys know right away that they need to change their pattern. Instead what you want is for them to first waste some time before they realize that something is wrong, and then waste more time experimenting on exactly what changed, and then even more time figuring out how to bypass it. So unless you're working on abuse at one of these companies, I don't understand how you can so confidently talk about what they are and aren't doing.
ravenstine 9y ago
Could that be a sign that current AI is not as intellgient as tech blogs suggest?
- tzakrajs 9y ago
  And that AI is a bubble within a bubble?
dhimes 9y ago
it seems like an existential risk to them Maybe the problem is that they don't think this is true. How will it put them out of business? Lawsuits? Government forcing them to pull the plug? People not using the platform because fake stuff is there? I'm not convinced that the end result of any one of these paths will be strong enough to wreck their businesses.
- azemetre 9y ago
  If people realize that not many users are actively using these platforms it would decrease the value of ad buys. I mean if 20% of all Twitter users are bots that would be detrimental to the sales department. If Twitter knew how many users are bots and did nothing it might be considered a form of fraud and deceiving investors. Why should I buy ads on a platform that are overpriced if the ads will not be seen by legitimate users?
  - dhimes 9y ago
    I think legitimate users will see them, but so will bots or fake users. I think there is a lot of click fraud, and has been since the early days of ppc. But we're still here. Now, I can't really answer your question, which was "Why" should you still buy ads on the platforms? I for one don't. But many, many people do. Just like they buy ads on TV knowing a lot of people are going to ffw through them.
lithos 9y ago
I listened in online to the intelligence and judicial committees. Kennedy and Cruz were pretty excited to bash Facebook's head with their oversights. And are probably hoping to get Facebook and Google both back in front of them. A bunch of other members spent time trying to get all three to say/agree to needing to be regulated. They are already in the middle of a very expensive and shackleing threat. It's just playing games to come out less behind (even if that's just going to be increased lobbying).
anigbrowl 9y ago
The market prioritizes short-term results over long term public goods. Think about it in microeconomic terms, aiming for a simplistic price equilibrium will give you the lowest common denominator outcome. The premise of the utility function selects for crap.

jsnell 9y ago

Just how do you know what these companies are using for fighting abuse (fake accounts, spam, account hijacking, etc)? Dealing with abuse is an adversarial process, where both sides are trying to adapt to what the other is doing. It's not like most programming where you can come up with a provably good enough solution. It also means that time is a valuable resource. There's nothing to be gained by e.g. Twitter announcing that they're now using ML on the profile images to detect fake accounts. That'd let the bad guys know right away that they need to change their pattern. Instead what you want is for them to first waste some time before they realize that something is wrong, and then waste more time experimenting on exactly what changed, and then even more time figuring out how to bypass it. So unless you're working on abuse at one of these companies, I don't understand how you can so confidently talk about what they are and aren't doing.

ravenstine 9y ago

Could that be a sign that current AI is not as intellgient as tech blogs suggest?

tzakrajs 9y ago
And that AI is a bubble within a bubble?

dhimes 9y ago

it seems like an existential risk to them Maybe the problem is that they don't think this is true. How will it put them out of business? Lawsuits? Government forcing them to pull the plug? People not using the platform because fake stuff is there? I'm not convinced that the end result of any one of these paths will be strong enough to wreck their businesses.

azemetre 9y ago
If people realize that not many users are actively using these platforms it would decrease the value of ad buys. I mean if 20% of all Twitter users are bots that would be detrimental to the sales department. If Twitter knew how many users are bots and did nothing it might be considered a form of fraud and deceiving investors. Why should I buy ads on a platform that are overpriced if the ads will not be seen by legitimate users?
- dhimes 9y ago
  I think legitimate users will see them, but so will bots or fake users. I think there is a lot of click fraud, and has been since the early days of ppc. But we're still here. Now, I can't really answer your question, which was "Why" should you still buy ads on the platforms? I for one don't. But many, many people do. Just like they buy ads on TV knowing a lot of people are going to ffw through them.

lithos 9y ago

I listened in online to the intelligence and judicial committees. Kennedy and Cruz were pretty excited to bash Facebook's head with their oversights. And are probably hoping to get Facebook and Google both back in front of them. A bunch of other members spent time trying to get all three to say/agree to needing to be regulated. They are already in the middle of a very expensive and shackleing threat. It's just playing games to come out less behind (even if that's just going to be increased lobbying).

anigbrowl 9y ago

The market prioritizes short-term results over long term public goods. Think about it in microeconomic terms, aiming for a simplistic price equilibrium will give you the lowest common denominator outcome. The premise of the utility function selects for crap.