Algorithm designed to show people what they want shows people what they want.
Now, imagine you have the job to recommend content for them. I feel like not only would a compassionate person know to avoid these triggers "for the user's own good" (which I would find way too paternalistic), even a really competent human (or a truly powerful AI) that was optimizing for "what this user wants" would avoid triggering content as the customer probably is even able to express this want verbally.
The issue, though, is that if you happen to ascertain (whether by active analysis or more passive experimentation) "omg every time I show them this thing they immediately drop everything else they are doing and spend three hours watching similar videos!!! this is amazing: I have found the best possible thing to show them!!!" and act on that despite it not being what the user wants you are just being evil.
And that's Google: the people at Google that work on this software are evil. Google employees seem to just be selfishly optimizing for "engagement" (to the extent to which they succeed... honestly they aren't even all that great at it, but they do much better than we would actually want ;P) and will just blindly show people stuff that they themselves might actively wish to avoid but can't help but get sucked into... so Google can make more ad money.
In the case of these conspiracy theories, I will contend most people want to be informed. They are momentarily skeptical of something, which makes them likely to get turned on easily by these videos, but then it is Google that radicalizes them by realizing "omg when I show them more videos of this form they just keep digging into the rabbit hole!!! SO MUCH AD MONEY" and "ugh when I show them videos from the other side they get turned off and maybe even drop the whole issue and go off to do something else more productive... that sucks! :/".
The engineers at Google have thereby built a system that, sure, "shows you what you want", but it shows you the most twisted, myopic, and even cruel version of what you want in the name of making them ad revenue over all other potential benefits to the user. Sure, sometimes this can be rationalized as showing the user something they legitimately found fun or interesting, and if you go into it with a careful mind and a wall of defenses it is possible to get what you need despite this perverted optimizer... but, I would hope we all can understand that it goes much deeper than that and that the negative effects are not ones that should be ignored in the name of making more money.
No company is required to provide intellectually healthy results. How would those be defined in a fair way? Is there even a good way of detecting less direct codeword driven searches that might silo perspective?
A better approach might involve multiple, hopefully not for profit, fact checking and news organizations. Each might have a Truth / Lies and Bias / Informative score among others that helps show the quality of the information. This metadata could be a first class citizen on the content. A warning in front of content that has not yet been vetted or which includes lies / large bias might help warn and inform end users.
Correct: it is the humans that built the machines in this way and continue to operate them that are malicious, not the machines. In this way I will note that it isn't even clear to me that companies can have malice: humans--employees of companies and developers of machines--have malice.
> No company is required to provide intellectually healthy results.
Correct: no one is required to not be evil. Google for a long time claimed to not be evil, but the people who work for Google are, in fact, quite evil :(. Regardless, the issue at hand was whether these systems are--and this is an exact quote from the post I am replying to--"designed to show people what they want"... they are not.
> And that's Google: the people at Google that work on this software are evil.
This is an asinine assertion. The people at Google that work on this are the same as you and I, motivated by precisely the same incentives. They're no different from the people at Microsoft or Amazon or Apple or Netflix, and many of them probably worked at those companies previously. The upper management working at any modern tech company will readily abuse their position as a service provider if it makes them more money. Apple does it with the App Store. Microsoft does it with Windows APIs. Amazon does it with Prime and their shipping networks.
The lack of regulation and blind money-chasing is what allows for evil decisions. There's no minion of Sauron working at Google, plotting ways to make your user experience subtly worse. Accusing them of that is ridiculous, and I hope you don't look at your coworkers the same way.
I have on multiple occasions been given the opportunity to make a lot of money by doing the things that people at companies like Google and Apple do. I have also on numerous occasions been given the opportunity to be hired by these companies to work against the interests I've been engaged in in the community. I have watched on in horror while friends of mine--people I thought I knew--chose to work on things that I consider evil and then suddenly start defending what they with the talking points we'd been arguing against, together, for years, proving they were never much more than a mercenary willing to work for whomever had enough money--or at least a fun problem to solve--rather than picking and choosing their opportunities to do good in the world.
> The upper management working at any modern tech company will readily abuse their position as a service provider if it makes them more money. Apple does it with the App Store. Microsoft does it with Windows APIs. Amazon does it with Prime and their shipping networks.
This is irrelevant. Are you seriously trying to say "the people at Google aren't evil because then we'd have to claim the people at Apple and Amazon are also evil"? If any assertion here is "asinine" here I'd claim it is that one :/. Or maybe you somehow think you are trapping me in a hypocrisy? :( if so, maybe you should check who I am before attempting such ;P. I mean, I'm literally currently suing Apple over that App Store monopoly, and the vast majority of my comments on every single media platform for over a decade has called out people--including the people I am talking about in the previous paragraph here--who work at Apple!
> The lack of regulation and blind money-chasing is what allows for evil decisions. There's no minion of Sauron working at Google, plotting ways to make your user experience subtly worse.
There is no such thing as regulation that can regulate away all potential for evil. As for "blind money-chasing": yes, that's evil. And I never claimed that the people at Google were "plotting ways to make [our] user experience subtly worse"... I claimed they are blindly optimizing for money despite knowing the ill effects, and that makes them evil. You often have an opportunity to do good in a day as a human, and often that decision to do good comes at the cost of some money you could have made by being evil... I take it you don't think, unless there is a regulation preventing it, anyone should ever be good, and we should never hold other people to the standard of being good? :(
> Accusing them of that is ridiculous, and I hope you don't look at your coworkers the same way.
The company I work at has an internal culture of calling people out when they get greedy, and we try to screen the people we hire not merely on skills but on viewpoint and ethics. This is sometimes a tense discussion, and we've had somewhat brutal arguments and even people (once even me! ;P) who have quit to make a point about various issues; but, in the end, I have been fairly confident that we have acted in an ethical manner and have frankly left a LOT of money on the table in so doing. So, no: I don't look at my coworkers the same way... if I did, that would probably be a sign that I should quit the company I'm working at as maybe we are doing something evil :(.
So, no: I do not at all believe this is "an asinine assertion".
FWIW, I would be willing to accept a very different argument from the one you are making here, wherein some people simply don't have the privilege of working on things that are good, as they are in a situation where they are living day to day under the boot of a system that requires them to do something, and maybe it is better than someone who feels a bit bad about it but is forced to do it anyway does the job rather than someone else, as it will have some imprint of that belief left... but most of the people I know who do these jobs just don't care: they find doing this kind of work fun and they try to not pay attention to the negative effects while making a ton of money doing but one thing out of myriad they could be doing with their often over-qualified skillsets... they simply find the thought process of taking a pay cut to work on better things beneath them.
I think the reason Google (and Amazon, and Netflix, and every other major tech company) uses clicks/view time as recommendation engine inputs is because... well, what else can they use? What quantifiable metric could possibly be used for large-scale, automated recommendations that more accurately indicates what someone actually wants to see more of? (This isn't rhetorical: if you have any ideas, I'd love to hear them.)
I don't think clicks/view time are the best metrics at all, and I don't think they're extremely accurate. But I also think they're the most accurate measure we've got, with the only other options being either (a) remove recommendations entirely, or (b) have humans manually monitor everything you do on the website, occasionally ask you why you clicked or watched things, and then make personal recommendations to you based on your answers. The latter of which is slow, more expensive, less scalable, more invasive to the user, and more tedious for the employees that would have to sit there monitoring you.
Maybe one day we'll have an AI method of using your comments and search terms as a better indication of desire (I dare say some of the recent LLMs are close to being ready for that task), but we're not there yet.
In the case of social networks, I think they served a positive function to both society and the people who used them back when they didn't have the recommendation algorithms, and your feeds were curated by you choosing to follow people explicitly; this, however, was not profitable, so we are now here.
If you are to do it, then yes: I think you probably need to do what TikTok is doing, and have humans heavily involved in the recommendation system in a way that attempts to put a hand on the wheel rather than the Google way to approach problems with algorithms on top of algorithms and no humans anywhere.
In similar vein, would there be a difference in algorithm that shows you what you want, or determines which videos create the most dopamine hits and juices you in a particular direction?
There may be fancier ways to describe this, but would an algorithm that merely enhances diffusion be the same as an algorithm designed to drive molecules to the most extreme directions?
> In similar vein, would there be a difference in algorithm that shows you what you want, or determines which videos create the most dopamine hits and juices you in a particular direction?
Yes, and the idea that you don't see this is literally shocking to me... it absolutely is the case that people can want things that they have a hard time avoiding, and the existence of a "dopamine hit" is not some kind of "proof of want". The comment I was replying to said that the algorithm was "designed to show people what they want"... that is categorically not true: it was designed to show people what makes Google the most money, full stop.
> There may be fancier ways to describe this, but would an algorithm that merely enhances diffusion be the same as an algorithm designed to drive molecules to the most extreme directions?
I'm sorry, but are you seriously now advocating for treating users--sentient beings with wants and motivations--as nothing more than molecules you can happily and gleefully move with your algorithms? Do you really not see anything wrong with this? :(
I happen to agree with you, actually. It's obvious to me there is a distinction between a preference-neutral recommendation algorithm vs. one juiced for corporate/profit interests. That being said, I don't appreciate personal attacks in replies and I don't think there will be any more useful comments on this matter at least on my part.