Good for Google, Bad for America | Hacker News Reader

Good for Google, Bad for America

nytimes.com

95 pointsAndrewBissell80 comments

Eric Schmidt chairs the Defense Innovation Board (DIB) [1], the US government knows about the Google/IBM China relationship.

One big concern is with the China/Google/IBM OpenPOWER [2,3] collaboration. Advanced hardware chip technology is one big thing China currently does not have.

The US doesn't want the Google/IBM OpenPOWER Foundation to hand China the keys to the kingdom in the form of advanced microprocessor chip technology and know-how, thus giving away the US hardware chip advantage, i.e. gifting the gap...

~> IBM Venture with China Stirs Concerns (2015) https://www.nytimes.com/2015/04/20/business/ibm-project-in-c...

~> China has never had a real chip industry. Making AI chips could change that (2018) https://www.technologyreview.com/s/612569/china-has-never-ha...

[1] https://innovation.defense.gov

[2] https://openpowerfoundation.org

[3] https://en.wikipedia.org/wiki/OpenPOWER_Foundation

China arguably is even further behind in software.

So we should make sure that all open source software development is shut down, especially since software’s role in military technology s only growing, lest China gets access to it.

I’m all in favor of Gothub being banned, and while we’re at it let’s get rid of the AOSP as well. And Linux is clearly a curse and while we’re at it let’s jail Microsoft execs for open sourcing .Net as well.

Some things are more significant than others. Like giving someone fire for the first time. Epochs mark the era before and after. Gifting the gap changes the game.

'At its core, artificial intelligence is a military technology.'

False premise. How is it a military technology at its core as opposed to any other technology? Or every technology is a military technology at its core, in which case, so what?

>A.I. is a military technology. Forget the sci-fi fantasy (...) these tools are nevertheless valuable to any army.

I guess cars and packaged food are also ‚a military technology‘?

Yes, especially when other armies were still on horses, food was fast to spoil and foodlines were hard to maintain. Planes were too, so were satellites. Still are. Any sufficiently advanced technology gives you an advantage and serves as a deterrent, especially when no one else has it. Giving it away is giving away your country and your allies' hard-earned, time-tested, long-term advantage. But that's just part of it, there are other factors to consider too. Huge unknowns on the horizon. The question is what is driving this play?

He owns Palantir.

Guess what they do....

I am dumbfounded at how many people are missing his point.

- When airplanes came out, their first application was military.

- When networks and computers came out, their first application was military.

- When satellites were first developed, their first application was military.

A.I. has a lot of uses, but it is another new technology, a whole new field of computer science that is rapidly evolving and has clear, identifiable military applications. And its early stage of development means that each breakthrough can pose a strategic military advantage.

This is true whether you agree with his point about the morality of Google partnering with China.

New medicines also find their way into militaries. Should we hoard those?

What about agricultural technology? Modern farming methods feed a billion people, but armies march on their stomachs. Should they all go hungry?

Funny you bring up the meds thing.

One of the problems that just surfaced in the news yesterday is the realization that almost all prescription meds are manufactured in China. The risk that poses to the US population and DoD is incalculable. How do we get off China meds?

I note you avoided the question.

In your view, should we hoard medical technology? Doing so will lead to countless unnecessary deaths and tremendous suffering. Is that worth it in your eyes to preserve a military advantage?

It is also clear that not every technology is permanently hoarded or seen exclusively as a military advantage.

If the Chinese were known to have stockpiles of Mustard Gas (a bio weapon not otherwise applicable to civilian life), then perhaps the US having exclusive knowledge of a cure or preventative substance would be a military advantage.

Your premise is that AI at its current stage is equivalent to a cure for cancer. I disagree.

A few points.

First, mustard gas is not a bioweapon. It's a chemical agent. There will never be a cure because it's like saying that you have a cure for fire.

Second, the point raised above is that things like food preservation technology implicitly confer military advantage. The same is true of many lifesaving medical technologies, especially things like celox that treat trauma.

And then there's cryptography. I find it hard to believe that you (or anyone on HN) support Joe Biden's approach to restrictions there. But that's quite clearly a dual use technology.

Finally, I never said it was equivalent to a cure for cancer. I don't believe that, and it's disingenuous to put words in my mouth. But things like self driving cars probably will actually save lives, and even independent of that I think it's worth examining very critically what restrictions we as a free society are willing to impose to gain military advantage.

No. Med tech should be as abundant as can be. But with AI there too many unknowns we're still figuring out. How to defend networks against AI attacks, hardening in general, how to handle the containment problem, the proliferation of deep fakes, deep truth and decerning between the two. And what that means at different levels. What are the critical points? What's the optimal path?

Dropping the hyperbole for a second.

Is there a line for ML (what we're erroneously calling AI here) where you think sharing it is appropriate? Self driving car tech could certainly save a great many lives; is it more reasonable to share the next Viagra than to save those people?

Selling them cars and chips is fine, to a degree. That's different than transferring IP and giving them independent capability because you can stop selling to them if need be.

Not sure if you've ever been to China, but suffice to say that IP law is really not a major issue there. Reverse engineering ICs is very common, and basically the only thing really stopping that from happening for top end ICs is that RE time eats into profits.

Yes, fake chips have been a problem too, polluting the supply space. All the more reason to protect and verify IP. There's new tech for that. Maybe the next big thing. Hardware verified. Authenticity.

The problem for your argument isn't the parts that don't work, it's the ones that do-- and there are plenty of them.

You keep seeing the part while missing the whole. Kick your perspective up a level, and maybe you'll put it together.

You have to apply some judgement; otherwise you reach ridiculous false equivalences. AI is clearly something interesting to both militaries as a potential strategic advantage.

> You have to apply some judgement; otherwise you reach ridiculous false equivalences.

An article about 'AI' that requires actual 'I' to understand? Inconceivable!

Wait until he finds out that Google researchers publish scientific papers about the latest machine learning advances for the entire world to read. And so do universities! Heck, teenagers are learning about this stuff! Why allow that?

The logical conclusion of "we shouldn't let China have machine learning" is that it should be locked down like nuclear weapons research. But it would be even more difficult to control, because you only need graphics cards, not uranium.

We cannot know what research they don't publish. Google also patents DL stuff, sometimes ridiculously broad like the patent for Dropout. And they gobble up any companies that have a chance to rival them, like DeepMind. Google is an advertising company though, it's where they make their money. I doubt we ll see real AI or weapons from them.

I find it more than a little surprising that no one here has an issue with google starting an AI lab in China. Instead all I see pointed out are the motives of the author.

The screwdriver is a military technology. Forget the sci-fi fantasy; what is powerful about actually existing screwdrivers is its application to relatively mundane tasks like driving screws and carpentry. Though less uncanny than Frankenstein’s monster, these tools are nevertheless valuable to any army.

i suppose you wanted nuclear research to happen in germany and japan rather than america then?

it’s just like a screwdriver, that’s why china is using AI for tracking and oppressing their uighur minorities

> it’s just like a screwdriver, that’s why china is using AI for tracking and oppressing their uighur minorities

I don't think a screwdriver can be used to oppress millions of people.

I think using screwdrivers as an analogy underplays the potential importance of AI. Electricity would be a better analogy.

nukes is the best.

The premises here range from true to arguable.

1) AI/ML is a dual use technology, one with both civilian and military uses...plausible. Worth noting that ITAR* already restricts transfer of a variety of technologies to others. Maybe AI/ML should count? 2) China is the enemy of the US...debatable 3) Google's presence in China gives it government greater access to these tech's than it otherwise would have...plausible

Seems like a generic opinion piece to me.

* https://en.wikipedia.org/wiki/International_Traffic_in_Arms_...

> 3) Google's presence in China gives it government greater access to these tech's than it otherwise would have...plausible

Even if Google didn't have an office in China, Chinese researchers could just as well learn from NIPS proceedings, etc.

This is not about NIPS proceedings. It's about advanced nextgen chip hardware capability, something China does not have. It would be like giving someone a rocketship when they don't know how to build and pilot a plane. Inorganic technology transfers -- phase shifts like this -- are ripe for disaster. Without internal time-tested checks and balances, gifting the gap is a recipe for chaos -- NOT a recipe for containment. How long will the chaos last? Hard to see. Does anyone have a model for that? It could lead to their undoing, and others' too. Haven't we learned that lesson yet?

Lots of noise about who Thiel is, not a blip on why he is wrong. Usually a good bet for him being right

you know someone’s right when people attack them as a person instead of their message

Should nations regulate AI like they regulate advanced military technologies? Or like they do crypto?

I say no, the genie is way too far out of the bottle for this to be effective. Any nation that wants an advantage in AI will have to secretly develop their own advances and rely on secrecy rather than regulation to preserve their advantage.

Any nation that attempts to regulate AI tech for military purposes will only be hurting themselves, and gaining little.

Hardware asymmetry is one of the things that secures your advantage. 10th generation chips vs 5th generation chips gives you asymmetric advantage, just like 5th generation fighters vs 3rd generation fighters are no contest. The ability to design and manufacture your own chips guarantees supply and the control to optimize and scale production capacity as demand requires.

We invented the microprocessor and SV has 40+ years of accumulated knowledge and hard-earned experience built into every chip. When an adversary is 40 years behind and has not yet developed the capability to make their own chips, that's significant. We rode that curve from beginning to end, and catching a ride on an exponential rocketship gives you a massive a head start -- a secure long-term advantage, one that's almost insurmountable if you don't give it away. So let's not do that. There's an optimal path. Let's find it.

China has extensive fab capacity down to 28nm and in-country access to processes as low as 16nm. That puts them roughly where we were in 2011. They also have market access to 10nm and will likely have market access to 3nm before the US has in-country access to it. Furthermore, the actual location of those advanced fabs are all in SE Asia, particularly Taiwan and South Korea. Tech transfer is inevitable, market access will continue to exist in peacetime, and in the event of any conflict they will simply take what they want. Meanwhile, the US is basically stalled around 7nm and has been for quite some time.

Taken together that means that we have at most an 8 year lead on what is in play literally right now in china. Given their investment level it's pretty likely that they will hit parity within two years. And Huawei was first to market with a 7nm CPU design. This is what you term a "secure long-term advantage"?

You watching the news?

Is there something on that dragged your attention away from the fact that your earlier assertions about the US's continued dominance of semiconductor fab were incorrect?

You mapped out a trajectory you think things are going to follow. Things change. No more technology transfer. Protecting IP is the key to what happened yesterday.

Things do change. That doesn't make your specific prediction of what they will change into correct. Given that your previous assertions about the past and present were wrong, I doubt you are correct about the shape of the future.

Happy to bet on it, though.

You're referring to part of the process, I'm referring to the whole process -- i.e. full independent capability / self-sufficiency -- and currently China has zero capability to manufacture advanced chips independently [0] ...

  Huawei would love to source its chips and other components 
  purely from Chinese suppliers, but as of today it is unable 
  to do so ... There is no production line in China that uses 
  only equipment made in China, so it is very difficult to make 
  any chipsets without U.S. equipment.

[0] https://wccftech.com/china-is-still-multiple-generations-beh...

Malarky. By the same measure the US can't manufacture anything at all anymore because it can't source all the parts of any single supply chain from solely inside the US.

China has-- currently, not theoretically-- in country fab just a few years behind, and has been actively building plants at 10nm. We know they've already bought equipment for 7nm. So the idea that somehow the supply chain is holding them back is just wrong.

Not only that, but U.S. equipment doesn't lead here and hasn't for quite some time. Taiwanese equipment does. Are you really arguing that China does not and will not have access to Taiwanese Fab equipment? Because if so, you should know that TSMC is already moving fab to China.

We will see.

Let me see if I understand this right: New Zealand citizen and Facebook board member Peter Thiel criticizes UK-based DeepMind and its sister company Google for working with non-US countries on AI technology.

In addition, there's no mention that three of Thiel's key investments (Space X, Palantir, and Anduril) require contracts from the US government, which is run by a President who Thiel helped get elected and is very hostile towards China.

Google is Thiel's rival, this should be the lens through which everyone reading this piece reads it.

Also no mention of the fact that palantir and anduril may have their own list of ethical challenges.

It's a huge fail by the NYT that they failed to disclose Thiel's role as a Facebook board member, etc. This is no opinion piece, it's aiming to benefit the writer and his interests/investments.

Many people have eyes on this. This is not about an investment.

It absolutely is an opinion piece which is why the disclosure is not required. Why would someone write an op-ed that's not to their benefit?

Yup. The quality standards on opinion pieces, even at NYT has drastically decreased.

https://news.ycombinator.com/item?id=20593547

This is not a FB thing -- It's a technology transfer thing. An economic thing. A trade thing. A national security thing...

There's a ton more going on in the news and behind the scenes beyond the Thiel NYT post.

And why exactly from a national security POV is Mr Thiel (now on his third nationality) allowed to involved with Space X, Palantir, and Anduril.

Probably because New Zealand is not a military rival and China is, this is pretty obvious to people who aren't being disingenuous.

In a lot of cases, it falls under similar reasons to why the US brought thousands of German scientists during and after WW2 via Operation Paperclip [1] and others, to work in places such as NASA. It becomes a matter of compartmentalizing secrets, understanding the individual's drivers, considering the levels of risks and potential ways to mitigate it, etc.

[1] https://en.wikipedia.org/wiki/Operation_Paperclip

This is a reductionist ad hominem response that fails to address any of the points raised in TFA.

> Let me see if I understand this right: New Zealand citizen and Facebook board member Peter Thiel criticizes UK-based DeepMind and its sister company Google for working with non-US countries on AI technology.

No, you do not have that correct. He is not criticizing Google for working with "non-US countries". He is criticizing them for working with China specifically, which is known to spy on and steal technology from the US and then use it to oppress their citizens. At the same time Google is refusing to work with the US, you know the country where they actually live and work.

Ad hominem accusations should not be used to try to hide large conflicts of interests.

Arguments are valid or invalid on their own merits, regardless of the interests of the person making them. If Thiel made a claim that was false or that did not logically follow from the premises, that would be worth pointing out. Merely suggesting that an argument is in someone's interest does not mean it's wrong. Are people only allowed to make arguments that are against their own interests? That seems unreasonable.

Pretty unbelievable the times published this opinion piece by Thiel. Without any real evidence to back his claim and given Thiel's past, its hard for me to imagine this isn't him pushing a false narrative for some emotionally charged ulterior motive.

What claims do you see him putting forth "without any real evidence"? I don't see how it's at all off base to assume China might take a military interest in AI labs Google establishes in Beijing.

EDIT: Ah, I see that Thiel said elsewhere that China already carried out espionage inside Google and hasn't backed this up. That claim is absent from this op-ed but it may be CYA for those comments on his part.

Context: Thiel originally made this claim in a speech where he claimed Google is committing "treason." [1] Given his "evidence" that's obviously insane. A large fractions of large multinationals have offices in China. I assume the editors at the Times told him that was a non-starter so he toned it down a bit.

Nowhere in the article does he disclose that he, at one point, owned 10% of facebook, a top competitor of Google's.

Most ironically, Facebook announced they'd open an office in china over a year ago [2].

He makes unsubstantiated claims that technology coming from this lab will be useful in the military before all else etc. He neglects to mention Google's public intent for the office, which has nothing to do with the military [3].

[1] https://www.axios.com/peter-thiel-says-fbi-cia-should-probe-... [2] https://www.theverge.com/2018/7/24/17607992/facebook-allowed... [3] https://ai.google/research/join-us/beijing/

China currently does not have the capability to design and make advanced hardware microprocessor chips. The Google/IBM OpenPOWER Foundation is simultaneously researching nextgen chips and teaching China how to do so.

Giving China this capability, i.e. gifting the gap, will give China an advanced hardware capability they do not have and did not earn. It will undoubtedly be used by the Chinese military. The supercomputer labs is the primary target.

The AI race is already a big unknown, and the Chinese government internal checks and balances are inorganic and not yet tested. Remember the tenets, be slow to speak and be wary of unearned wisdom. What is the point of accelerating this?

>EDIT: Ah, I see that Thiel said elsewhere that China already carried out espionage inside Google and hasn't backed this up. That claim is absent from this op-ed but it may be CYA for those comments on his part.

I'm sure editors at the Times struck that as well. Still it makes you wonder, why publish him at all?

Google publicly announced in 2010 that China hacked them, and that's why Google withdrew from China in the first place.

was there any google entities linked to the HP cloud europe china hacker break ins?

I think we should all totally disregard Thiel as a lunatic. It's not as if he has been right in any of his choices so far

He was right about funding the Trump campaign.

I think there’s a difference between betting on a winning horse and making the right choices.

among other things. i forgot the /s in my comment

I would run the other way from any political campaign that Thiel finds agreeable. The man is an oligarch that doesn't care for democracy.

FWIW, I don’t take anything from the NYT (especially opeds) at face value. Same goes for generally any media outlet, but NYT (also other outlets) and tech companies are competitors. Keep that in mind and take the NYT’s weekly piece on Google, FB, Amazon, Apple etc almost being the worst companies in the world with a pinch of salt.

Good for Thiel, Bad for Journalism. I expect to see people pushing for their own agenda on Twitter, not on a journal that won 100+ Pulitzer prizes.

I fail to see how this is distinct from any other opinion article in the NY Times.

How many NYT op-eds has Thiel written? He's only written one tweet [1].

It's rare. The significance of that should tell you something.

[1] https://twitter.com/peterthiel

all that rag is used for is pushing people’s agenda. you just don’t notice it until it’s somethijg you disagree with

Reply on news.ycombinator.com