Tensions in Google's ethical AI group increase as it sends demands to CEO • Hacker News Board

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#311

Post by hn_throwaway_99 » Thu, Dec 17, 2020, 5:34 PM UTC

renewiltord wrote: post #201

hn_throwaway_99 wrote: post #117

Wow, Google's ethical AI group sounds like the most toxic group of individuals I've ever heard of. I'm going to sidestep the debate about whether Gebru was "fired" or she "resigned", because that debate primarily centers on how one defines the words. But clearly she did do this: She sent a letter that made a bunch of (IMO) outrageous demands, outrageous because it essentially attacked other people at the company, and…

This is just how you behave: when faced with opposing pressure you roll over. That's fine, I do the same in some situations: I don't fight people who can cause me great damage like police or air stewards or immigration agents. But rolling over all the time is not an optimal strategy. The more power you have, the rarer you should roll over. And of course many without power will see those with power and say "In their p…

> So being a pushover comes naturally to you, you've optimized well.

Talk about damning with faint praise. While my mind is honestly saying "seriously, fuck off", I'll try to put this in a more constructive response.

It's not about "rolling over". It's about having the maturity to realize when you are dealing with management, if by your demands you put them in an absolute no-win situation, then you've only screwed over yourself. That no-win situation was the demands by Gebru and the ethical AI team that management throw their colleagues under the bus. Management simply can't accede to that demand without saying to a huge portion of their employees "We're going to trash the reputations of other employees based on who yells the loudest".

Gebru and the ethical AI team most probably could have gotten huge portions of what they wanted (perhaps not all, but most), and especially would have better actually implemented the organizational change they say they want, if they hadn't handled the situation so extremely unprofessionally.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#312

Post by bonoboTP » Thu, Dec 17, 2020, 5:34 PM UTC

pietrovismara wrote: post #293

igorkraw wrote: post #282

I find it sad that a good chunk of HN seems to just go with the "she's a toxic person" narrative without ever relaying what she actually DID. Ironically, the self identified "anti-SJW" are cancelling her by complaining about her "toxicity" and "attempts to cancel people" in vague terms. I also find it sad that apparently, freedom of speech and free actions end when it comes to criticizing your employer in public, sho…

We grow up with the ideal of democracy and treat it as the single most important value we have to protect. Yet we spend the best (awake) third of our life in authocratic environments, where decisions that can deeply affect our lives are taken without our consent. Where speaking up for ourselves or our peers' good gets us fired. I also find it incredibly sad.

The division is metaphysical / epistemological / narrative / tribal / ideological / religious etc. Way deeper than who is right in this Gebru vs Google debacle or the specifics of this one paper.

These debates go nowhere because both sides see the other as toxic and a threat and don't agree on basic terms, the basic view of who is in power, who needs to fear for their jobs for dissent etc.

"but here Gebru was fired, so POC and activists have to fear" "but Damore was also fired for dissent" "but he was sexist and created a toxic work environment" "but Gebru uses the race card for minor reasons if she doesn't get it her way" "now you are just racist and consider her an angry black woman" "I don't care about race just individuals" "comfortable from your place of privilege to keep the status quo, isn't it?" "I just want to do my job in peace without drama" "silence is violence when the system is structurally racist!" etc etc.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#313

Post by bisectable » Thu, Dec 17, 2020, 5:34 PM UTC

DoreenMichele wrote: post #49

shoo wrote: post #30

One thing I've wondered: what is the purpose of Google's ethical AI group? What utility does it provide to the corporation? I suspect the primary motivation for funding ethical AI work is _public relations_ (or less kindly, corporate propaganda) - to help mitigate damage to Google's brands and reputation by public concern surrounding artificial intelligence. It is well understood that it can be difficult to get someo…

Without actually knowing the history, I think it's probably stupidly conceived. We seem to generally imagine that the biggest problem or challenge with AI is ethics because we frame ethics as being largely rooted in something emotional, like empathy or compassion, and not logic. The movie I, Robot explores what happens when "logic" is taken to its logical outcome by an AI. It posits a robot with feelings as the antid…

Almost there. Each individual comes with a distinct set of biases. Social collective action is the contextual weighted average of those individual biases. This results in quite resilient societies: at equilibrium individual biases balance each other across the population, small changes in a relatively small number of individual preferences result in societal course corrections, avoiding dangerous shocks. Smooth ride.

A subtle AI risk is simply the centralization of the decision making process. This leads to the reinforcement of some subset of biases, until they completely overshadow alternative options. The system becomes rigid, having lost its day-to-day individual level feedback, thus its capacity to adapt to new circumstances. Sooner or later, it will fail.

See also 'too big to fail' and 'central planning'.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#314

Post by bo1024 » Thu, Dec 17, 2020, 5:36 PM UTC

InsomniacL wrote: post #142

mooneater wrote: post #37

> accepted her resignation immediately In saying "resignation", CNN is presenting the management framing -- as many people have pointed out, she was fired. Bloomberg states both sides of that: https://www.bloomberg.com/news/articles/2020-12-16/google-ai... As does The Verge: https://www.theverge.com/2020/12/7/22158501/timnit-gebru-tea...

"No one told me that I was fired." "Thanks for making your conditions clear. We cannot agree to #1 and #2 as you are requesting. We respect your decision to leave Google as a result, and we are accepting your resignation." https://twitter.com/timnitGebru/status/1334364734418726912

That's not quite the whole story. She set a deadline (a few weeks?) after which she would resign, and Google terminated her immediately that day.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#315

Post by istjohn » Thu, Dec 17, 2020, 5:37 PM UTC

jjcon wrote: post #129

js8 wrote: post #114

Earlier quoted context omitted.

> I understand that some may be upset about things their employer does and that's well within their rights. It is within their rights, true. But it's not just a question of law, but also ethics. We accept that, in democratic societies, we can freely criticize our government and still stay the citizens. There are good principal and practical reasons to have it so. However, in private companies, for some reason, we don…

I think the primary reason for this is because people (generally speaking) in democratic societies share control of their government (for the people by the people etc). This is not true of company->employee relationships. It is assumed that if you join a company you share their values and if you no longer share their values the barrier to leave and join another is quite low. People have different values and that is f…

Seems to me you have an overly simplistic view of what a corporation is. A corporation is not merely its stockholders. A corporation is accountable to multiple stakeholders, and it's stockholders are just one of many such stakeholders. A corporation's employees and customers are also stakeholders, as well as the society and nation in which it is embedded. It is perfectly legitimate for any of the various stakeholders to attempt to influence the behavior of the corporation.

If you are for tolerance and the coexistance of different values in society, you will understand that values come into conflict constantly. And you will respect the prerogative of employees and customers to push back when their values conflict with corporation management (who represent stockholders).

When you insist that employees must unilaterally disarm by walking away at the first sign of conflict with their employers, you are essentially expressing a class preference. You are asserting that the working class must submit to the stockholding upper class. A corporation requires capital to compete in our economy. And when the top 10% own 80% of the wealth, workers cannot simply walk away.

Stockholders can't walk away either. They need workers, just as workers need capital. So they will need to find a way to coexist, but that doesn't mean there won't be conflict. There will always be conflict. The virtue of our democracy is that the conflict is manifested as demand letters and PR statements instead of clubs and bullets.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#316

Post by pietrovismara » Thu, Dec 17, 2020, 5:41 PM UTC

bonoboTP wrote: post #312

pietrovismara wrote: post #293

Earlier quoted context omitted.

We grow up with the ideal of democracy and treat it as the single most important value we have to protect. Yet we spend the best (awake) third of our life in authocratic environments, where decisions that can deeply affect our lives are taken without our consent. Where speaking up for ourselves or our peers' good gets us fired. I also find it incredibly sad.

The division is metaphysical / epistemological / narrative / tribal / ideological / religious etc. Way deeper than who is right in this Gebru vs Google debacle or the specifics of this one paper. These debates go nowhere because both sides see the other as toxic and a threat and don't agree on basic terms, the basic view of who is in power, who needs to fear for their jobs for dissent etc. "but here Gebru was fired,…

This is true, but perspective is very important.

For me realizing that I'm not actually living in a democracy was what gave me an entirely new perspective on how the world works. It took me almost 30 years to realize this, so I know it's not really obvious given how we've been taught to think.

That's why I try to spread this message when it fits with the topic. And I feel so many problems have this issue at their core, including this one.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#317

Post by dang » Thu, Dec 17, 2020, 5:42 PM UTC

"Disingenuous" implies that someone is being deceptive, which breaks the HN guideline against assuming good faith: https://news.ycombinator.com/newsguidelines.html. The rest of your comment would be just fine without that bit.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#318

Post by matthewdgreen » Thu, Dec 17, 2020, 5:42 PM UTC

lawnchair_larry wrote: post #286

matthewdgreen wrote: post #162

Earlier quoted context omitted.

For Google, a huge percentage of the value in having an ethical AI division comes from having a credible and independent collection of experts who oversee the company's efforts. The minute that team ceases to be viewed as independent and credible, their value to Google is vastly diminished. What Google wants to say to the world is "hey, you should totally trust us, we have the best ethical AI people in the field look…

Is there any chance evidence to suggest that they hired Timnit merely as a PR show, rather than because they actually wanted a good product? Although it’s fun to assume bad faith when it comes to every action by every company, she was not supposed to be an independent critic to launder evil. She was supposed to work in cooperation with them to make a high quality product. They don’t want to create something that caus…

> Is there any chance evidence to suggest that they hired Timnit merely as a PR show, rather than because they actually wanted a good product?

I didn't mean to imply that the ethical AI team is empty PR. What I'm saying is that the obvious corporate justification for an (expensive!) AI ethics team is to reassure regulators and customers that Google's AI techniques will follow best ethical practices. Obviously the best way for Google to achieve that goal is to hire a bunch of really strong researchers, and then incorporate their recommendations into Google's products. And of course, to have those researchers publish their findings so that the world can see that a standard is being followed.

But the flipside can also be true, and damaging the credibility of your ethical AI team comes at a cost. It would not surprise me if some observers have come to negative conclusions about the independence of Google's researchers, which means those people are more likely to mistrust Google's products in the future. The irony is that even if Google has the very best intentions, they might now find it harder to convince people of this. This does meaningful harm to Google, and nothing Google's gotten out of this mess has been worth that cost.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#319

Post by frongpik » Thu, Dec 17, 2020, 5:43 PM UTC

optimalsolver wrote: post #262

frongpik wrote: post #189

A big reorg is coming to that ethics group.

And the new group will happily sign off on any future Google actions as completely ethical.

Perhaps the new group won't confuse ethics with activism.

Re: Tensions in Google's ethical AI group increase as it sends demands to CEO

#320

Post by Udik » Thu, Dec 17, 2020, 5:43 PM UTC

uh_uh wrote: post #247

AlanYx wrote: post #198

Earlier quoted context omitted.

That lawsuit threat is really a critical contextual element here. Having a manager who's openly threatening to sue the same company she acts as a manager for creates a host of complicated legal issues for the corporation.

It honestly baffles me. I also expect had I - or any of the people I know - made similar threats, we would be fired on very short notice. The fact that Google just let this past threat slide seems highly unusual to me.

It just makes you wonder how much toxic behaviour did her managers put up with before deciding that enough is enough, and (more or less) firing a prominent, visible, black, female AI ethicist working in the field of gender equality and racial discrimination. They knew perfectly well how damaging it could have been and yet decided they had enough.