AI at Google: our principles • Hacker News Board

Re: AI at Google: our principles

#391

Post by fwip » Fri, Jun 08, 2018, 6:44 PM UTC

LambdaComplex wrote: post #244

ergothus wrote: post #39

Earlier quoted context omitted.

FWIW, since they updated the google news algorithm, I've started getting a pretty diverse set of results. It might be because it picked up that I try to read opposing views, or it might be because the algorithm is terrible, or any other number of reasons, but regardless of the source it's refreshing to see. ...and revealing! The headlines differences between, say, Washington Post, CNN, The Hill, and Fox News for the…

>The headlines differences between, say, Washington Post, CNN, The Hill, and Fox News for the same news blurb is even more dramatic than I'd expect when you put them up side-by-side. Can you give a concrete example of this? I'm definitely interested in seeing how big the differences are

An example from today (first political result on my screen):

- Reuters: Trump says Russia should be at G7 meeting, Moscow not so sure

- CNN: Trump: Russia should be in the G7 summit

- TheHill: An isolated Trump attends the G-7

- Fox News: Trump prepares for North Korea summit as a great performer -- like Reagan

Re: AI at Google: our principles

#392

Post by dredmorbius » Fri, Jun 08, 2018, 7:04 PM UTC

kanox wrote: post #171

bobcostas55 wrote: post #37

>Avoid creating or reinforcing unfair bias. I recommend _The impossibility of “fairness”: a generalized impossibility result for decisions_[0] and _Inherent Trade-Offs in the Fair Determination of Risk Scores_[1] [0] https://arxiv.org/pdf/1707.01195.pdf [1] https://arxiv.org/pdf/1609.05807v1.pdf

> Avoid creating or reinforcing unfair bias. This very high up and is written in a way which would explicitly allow "fair bias". This means activists will have a free hand to use their positions at Google to enforce their vision of political orthodoxy.

Bias and discrimination are foundational to information and decision systems. Thate's wwhat they do: decide, choose, and filter.

Pichai's point is that such discrimination must be fair and societally beneficiaĺ.

The line is not clean, straight, or constant. But it provides a guiding principle to future decisionmakers aand stakeholders.

Re: AI at Google: our principles

#393

Post by Daishiman » Fri, Jun 08, 2018, 7:08 PM UTC

rmk wrote: post #381

brandonsometig wrote: post #337

Earlier quoted context omitted.

> Which country came to the aid of Britain, and all those Western European nations when they were faced with the existential threat of Nazism? Russia mainly, from where an estimated 24 million people died. American also didn't wade into a war through purely selfless means either, if Germany managed to invade Britain and western Europe, America would have been at significant threat.

India (part of the British Empire) also sent millions who died fighting in North Africa. The reasons every country got into the war were complex. But it is also largely true that America spilled blood and treasure out of a sense of obligation to fight Nazism, despite having a significant immigrant German population (who fought against their brethren on the other side), going so far as to impose a draft in the later s…

This is so much BS. You should actually read the history of the reasons for the involvement of the US in the WW2 theaters.

You sound like a young kid who has never actually read about history and geopolitics.

Please dude, actually read up on the involvement of the US in the secret wars in Cambodia by Kissinger, the School of the Americas, instrumental in teaching South American militaries repressive strategies that killed tens of thousands of innocents with the full support of the American government, the Iran-Contra affair, the ridiculous involvement in Vietnam.

There is an innumerably long list of atrocities commited purposely by the American government and with the silent consent of the American majority.

Seriously bro, Irak happened less than 15 years ago in a completely manufactured war and you're actually so stupid to believe that there's anything particularly worthy of American Imperalism?

You should try to be on the receiving end of the American business interests that have fueled these conflicts, lest we see what your opinion on the matter would be.

Re: AI at Google: our principles

#394

Post by dredmorbius » Fri, Jun 08, 2018, 7:09 PM UTC

EpicEng wrote: post #55

So, I'm all for giving someone the benefit of the doubt if they have a change of heart upon reconsidering an issue, but this coming after the fact rings a bit hollow to me. I think the only principle at play here is that it became a PR issue. That's fine, but let's be honest about it. Early emails between Google execs framed this project only in terms of revenue and potential PR backlash. As far as we're aware, there…

There's a lot to be said for this view, and I've reshared and highlighted your comments elsewhere.

At the same time, Google and Alphabet Inc., far-from perfect entities which I have and do criticise strongly and often, at least present a public face of learning from mistakess rather than doubling down or issueing repeated and increasingly trite and vacant apologies (say: Facebook).

This is a line Google have drawn for themselves in the sand, and our various future selves may judge them against.

Re: AI at Google: our principles

#395

Post by zoltaan » Fri, Jun 08, 2018, 7:23 PM UTC

fixermark wrote: post #378

zoltaan wrote: post #322

Earlier quoted context omitted.

I don't get it. Do you suggest that Google or big players make things better than anyone else? Or other way around, small startups cannot make good products? I believe if it was so then only companies starting with 1000+ people could do anything useful, small startups were doomed to fail, also big player could never do wrong. Sounds a bit simplistic and contrary to countless examples from life.

In the specific space of machine learning in life-critical applications, I generally prefer product from established companies with reputation to sub-1000 startups. The incentives to cut corners and go to market are much higher for small startups with short runways. I don't want corners cut when lives are on the line.

"I don't want corners cut when lives are on the line"

...still not sure if you are talking about the (heavily) cost optimising conglomerates or not....if we agree to constrain the topic to finances leaving out innovative ideas, ethics, integrity, trustworthiness, etc. where conglomerates may be loose with standards.

Also do you know how Apple and Google started? (I hoped the suggestion will get through without stating the obvious, but it did not)

Re: AI at Google: our principles

#396

Post by kchr » Fri, Jun 08, 2018, 7:23 PM UTC

munificent wrote: post #234

hueving wrote: post #144

Earlier quoted context omitted.

This is a bit dramatic though because these robots can't reload themselves. Automated targeting for a single area is miles away from planning strategy, deploying battle groups, etc.

> This is a bit dramatic though because these robots can't reload themselves. Neither can mines, and look at the decades of devastation those have caused. Now imagine a minefield where the mines get up and chase after you.

Truly the stuff of nightmares. Makes me think about that episode of Black Mirror with the Big Dog (DARPA) inspired surveillance robot...

Re: AI at Google: our principles

#397

Post by simion314 » Fri, Jun 08, 2018, 7:29 PM UTC

RyanZAG wrote: post #386

simion314 wrote: post #384

Earlier quoted context omitted.

Sure, you are not forced, I said "CAN", Say US wants to spend a lot of many with some black magic consultant that could assassinate at a distance, you can justify it by launching a rumor that China does it too or probably does it or it will do it. So you throw away any moral discussions by blaming China , they do it so we have no choice.

Hmm makes sense. I see what you mean from the position of say a general in US army tasked with this stuff: if someone says China can do X, his bosses are going to ask why America can't do X too. And therefore to force America into doing X, you just need reliable sounding misinformation to prove China doing X, and then moral arguments are swept aside. However that aside, AI is very big in China right now, and they're…

Sure, I would like to see the real reason, We want drones to strike in 3rd world countries. Is not like China will send drones in US.

People don't like war, so it is natural that some people won't want to use their talent for making weapons.

Re: AI at Google: our principles

#398

Post by fesoliveira » Fri, Jun 08, 2018, 7:41 PM UTC

AsyncAwait wrote: post #349

themacguffinman wrote: post #65

Earlier quoted context omitted.

I doubt that Google spelling out their moral stance is intended to convince you right away that they're all good now. It's a public standard that they're setting for themselves. If you think their actions don't match their words, you now have concrete terms and principles to critique and compare with. It's a benchmark to which employees and the public can hold them accountable.

Even if they abide by this, who's to say that once a party has some Google developed military AI, they won't misuse it? I fail to see how Google can effectively prevent this.

If they develop an AI that administers medicine to veterans, and the army takes it and change it so it will administer torture substances to prisoners of war, is it Google's fault or the army's fault?

Google makes tools with a purpose in mind, but like many other technologies in history, they can always be twisted into something harmful, just like Einstein's Theory of Relativity was used as basis for the first nuclear weapons.

Re: AI at Google: our principles

#399

Post by VanillaCafe » Fri, Jun 08, 2018, 7:53 PM UTC

EpicEng wrote: post #55

So, I'm all for giving someone the benefit of the doubt if they have a change of heart upon reconsidering an issue, but this coming after the fact rings a bit hollow to me. I think the only principle at play here is that it became a PR issue. That's fine, but let's be honest about it. Early emails between Google execs framed this project only in terms of revenue and potential PR backlash. As far as we're aware, there…

These principles seem to have already been wholesomely applied to Google then Waymo's approach to self-driving cars. This is in stark contrast to the approaches used by competitors such as Uber and Tesla who appear to favor capturing market first and foremost.

It seems a narrow view to take to assume Google's only AI project with mortal human consequence is Maven, then using that narrow view to confirm your own negative bias about profit, perception, disingenuity.

Re: AI at Google: our principles

#400

Post by sunsetMurk » Fri, Jun 08, 2018, 7:57 PM UTC

repolfx wrote: post #345

ergothus wrote: post #39

Earlier quoted context omitted.

FWIW, since they updated the google news algorithm, I've started getting a pretty diverse set of results. It might be because it picked up that I try to read opposing views, or it might be because the algorithm is terrible, or any other number of reasons, but regardless of the source it's refreshing to see. ...and revealing! The headlines differences between, say, Washington Post, CNN, The Hill, and Fox News for the…

When did that happen? I'd given up on Google News, it reached a point where it only wanted to show me articles from the Guardian or CNN even though I read a wide range of outlets. I just went back and looked again and there does indeed seem to be a better mix of stories now, although I am still very skeptical - the News algorithm is even less documented than the web search ranking algorithms and it would be tough to…

for some reason the new Google News app needs is a net new download/install of the app.

https://itunes.apple.com/us/app/google-news/id459182288?mt=8

When they demo'd at GoogleIO I was pretty excited to try and then I went to look at my already installed google news app and it was still the same old one I'd been using for a while. Long story short, discovered it's a whole different app with the exact same name but updated icon.

The actual app is very very good though. I've been reading lots more news in the app. It's been about 3 weeks using daily and I'm starting to notice reinforcement of my common subjects & sources on the 'For You' page. I hope I start to notice more curve balls. It's great to the 'Headlines' page which I'm pretty sure is the same for everyone.