Some of the best and most impactful AI research comes from Google DeepMind or Facebook AI research. Those ad clicks are subsidizing a whole lot of scientific advancements.
You're not wrong, but that argument can kind of be used to justify almost anything.
"Yeah, what we're doing to make a ton of money right now is skeevy and not adding any actual value the world, maybe even creating negative value, but we promise we're going to spend all of that money on research to advance humanity."
Google and Facebook actually did end up doing that, to good effect, but 1) was it worth it? (maybe, especially if we get benevolent superintelligent AGI directly out of it and Google / FB actually agree to not monopolize or abuse it), but 2) how many times should we accept that excuse from companies? If in the next 10 years, a hot up-and-coming new startup begins to grow rapidly and looks like it'll end up in the FAANG tier, should we also trust them when they say their questionable tactics are worth it because they're gonna spend that money on cool science stuff?
What's the alternative? Increase government R&D spend to the same levels as FAANG companies? But will they have enough infrastructure to support that? For example, Tensorflow runs on Google's global datacenter infrastructure - this is the only way these latest monster transformer models can be trained, using thousands of GPUs or TPUs.
The alternative would be to create a tech or hard science company which doesn't choose advertising as their core business model. Many large and successful tech companies already do this, and are contributing a lot to ML research and other fields. It's just that Google, Facebook/Instagram, and Twitter, some of the largest companies in the space, are inextricably tied to that model.
Apple and Netflix don't share the same problem, and that gives them a lot more freedom in other areas, too (but perhaps resulting in them having less total user acquisition potential than Google or Facebook; but of course it's needed less, since every customer is directly giving them money). Uber also doesn't share that problem, and they've contributed quite a bit to the community, though of course they have other problems.
The main problem isn't tech companies or monopolies, IMO; it's the tech companies that survive only by cannibalizing both non-paying users and non-paying non-users who happen to visit just about any website on the Internet (because odds are any given website is using Google Analytics or has some sort of Facebook or Twitter integration). It creates bad incentives.
Of course, there are also separate debates to be had over general issues of monopolization and working conditions (like Amazon fulfillment workers not being able to take bathroom breaks without risking accumulating points which may result in them getting fired), and control over different media platforms (like Google owning the only real video platform in the world and Twitter owning the only real microblogging / "town square" platform, and issues that may come from how they regulate users and content). The advertising stuff is just the cherry on top.
I do think trying to regulate or break them up over the content policing stuff would be a huge mistake, though. Staunch conservatives say they want them to relax their standards and stop being biased against conservatives; staunch liberals say they helped the spread of disinformation, racism, and fascism and should have tighter standards. I think if either side has their way, to any degree, it will be a disaster.
The alternative would be to create a tech or hard science company which doesn't choose advertising as their core business model...It's just that Google, Facebook/Instagram, and Twitter, some of the largest companies in the space, are inextricably tied to that model.
The main problem isn't tech companies or monopolies, IMO; it's the tech companies that survive only by cannibalizing both non-paying users and non-paying non-users who happen to visit just about any website on the Internet (because odds are any given website is using Google Analytics or has some sort of Facebook or Twitter integration). It creates bad incentives.
Wouldn't the world be a better place if the Internet had true micropayments? Culture would no longer dance to the whims of advertisers and executives at huge companies. There would be a more direct connection between creatives and consumers. Then again, we've tried instantly collated online mob rule in the form of social media karma. The nearest equivalents, in the form Patreon supported Instagram models and YouTube stars like PewDiePie and Jake Paul, on the face of it, don't seem to point us in the the best direction.
A huge irony is that the public wants free stuff, and historically has railed against micropayments. However, while they complain, it would seem that microtransactions are going strong, though they are arguably exploitative.
I'm not sure how good of an idea micropayments are. If every single website and app you ever used required micropayments, using the Internet would just be annoying. You would be incentivized to use as few services as possible, for one.
They work for certain kinds of things, but I don't think we'll see the day where people are regularly sending money for the right to open a blog or create an email account.
I don't really know what the answer is for large scale apps with non-paying users. Hopefully other forms of monetization become more popular.
It wouldn't connect people across the borders anymore and would create firewalls across countries just based on money. Ironically even places like HN wouldn't work, sure for wealthy Americans, but not for poor Indian or Ukrainian hackers just about about to develop their interest in tech.
What do you mean? " (like Amazon fulfillment workers not being able to take bathroom breaks without risking accumulating points which may result in them getting fired)." that isn't true and it's very misleading for you to post. I am a full time Amazon employee and I use the restroom as many times as I want in a day without being questioned and what are these points you're speaking of about accumulating points? You don't get fired from things that simple, you get fired from write ups, misconduct behavior etc.
Governments and government research institutes already are building supercomputers for research in domains like physics or biology. Given the increasing relevance of AI, it's only a matter of time that they also build supercomputers that have AI accelerators.
Government spend can't create a Google. Google depends on huge compensation packages to pampered engineers who get sushi for lunch and Segways for entertainment. A government facility of this nature would be political suicide.
Historically, a lot of successful public research was funded by either defense or intelligence agencies. It's hard to escape from "doing skeevy things"
That's true, but they already started out skeevy. Everyone already knows that stuff is gonna be skeevy. There's no other way for militaries or intelligence agencies to work. It pretty much has to be that way.
But some new tech startup certainly doesn't have to be skeevy to make money or to be in a position to contribute to research.
> That's true, but they already started out skeevy.
That's not true. A lot of times, the projects will start before funding.
> But some new tech startup certainly doesn't have to be skeevy to make money or to be in a position to contribute to research.
That is true, but you're also missing the reality of startups. Just survival alone is hard. Stubbornly sticking to idealism doesn't help that cause, which is why people take money from both China and Saudi Arabia.
It wasn't "basic" research though, as the basic research in AI happened a few decades ago. Google is industrializing Hinton's work, but hinton created most of it in toronto.
That said, industry can definitely do basic research work (e.g. Pharma does). But what google is doing is acquiring any company that might be a real threat to them.
Are you suggesting Google doesn't have real academic style research groups?
Surely you jest.
Didn't they even crossover into biology on one paper? They're a profit minded entity yes, but no way is it all shaving nanoseconds off of click throughs.
The broader theory of Peter Thiel is that most of these advancements have been done in the rather unregulated ‘world of bits’, while areas related to physics, chemistry etc. have seen less advancement.
Playing Go against an AI opponent while in a 1970s, crumbling, subway system would make for a good analogy.
Sure, but Google also acquired Android, Apple acquired Siri, and Microsoft purchased DOS. They were still developed for years under their new companies. At some point, it does become their own project.
"Yeah, what we're doing to make a ton of money right now is skeevy and not adding any actual value the world, maybe even creating negative value, but we promise we're going to spend all of that money on research to advance humanity."
Google and Facebook actually did end up doing that, to good effect, but 1) was it worth it? (maybe, especially if we get benevolent superintelligent AGI directly out of it and Google / FB actually agree to not monopolize or abuse it), but 2) how many times should we accept that excuse from companies? If in the next 10 years, a hot up-and-coming new startup begins to grow rapidly and looks like it'll end up in the FAANG tier, should we also trust them when they say their questionable tactics are worth it because they're gonna spend that money on cool science stuff?