Back to drawing board as Google cans AI ethics council amid complaints over right-wing member Google has canned its AI ethics board after just a week due to outcry over its choice of panel members, claiming it would find "different ways" to bring in outside opinions. The Advanced Technology External Advisory Council (ATEAC) was announced on 26 March and immediately inundated with complaints over the inclusion of Kay …
Its job was to whitewash Google's lack of ethics by allowing Google to claim all the evil stuff they do was passed by their ethics board. They probably needed a diversity of opinion because they would likely require the board vote a supermajority or even be unanimous to stop Google from doing something on ethical grounds.
They needed a warmonger or two on the board so they could work with the DoD on Skynet's AI while claiming they were observing the highest ethical standards because their board approved it.
If you do not choose an authority for ethics, then you need a broad and diverse ethics board.
Thus you either pre-assume your conclusion that A is right and B is wrong, or you have a mix of A and B and hear each out.
So it's less an "ethics" board, and more a "enforcement board". That's fine. You just have to be honest. To yourself and to others. Are you learning about ethics, or imposing your ethics?
So will they try again to pick a new "Ethics advisory board" that only includes people with a specific viewpoint set, and if they accidentally let in some people representing opposing views, will they trash the next board and try again till they get one that all has the same specific viewpoint?
Our new AI overlords might just see that this explains why human input is unreliable, useless and decide to do what they want and who cares what the meat sacks think.
I'd honestly argue that having a diverse group on the council is better than having one chosen from the various Google staffer safe spaces where it will predominantly be left-leaning.
They need to get a range of people on there if it's to be useful, which I doubt it will be anyway. However there is a slim chance that it'll be more than a rubber stamping system for Google's random unethical AI plans.
predominantly be left-leaning.
Well as the modern right wing are almost by definition unethical, unless the exercise was to rubber stamp evil actions a leftward stance is inevitable.
Perhaps a reformation of the right wing away from persecution of non-heterosexuals, understanding of the urgency of acting to mitigate climate change and all the other nasty views they seem to enjoy would be a better idea.
Maybe not intrinsically but there certainly seems to be a high degree of correlation of socially abhorrent views and right wing tendencies.
I don't say there are no left leaning bigots, there are, but there are a lot more right leaning ones and they seem to embrace it as part of their political stance and are encouraged to do so by the actions of some right wing "leaders" such as Trump.
"high degree of correlation of socially abhorrent views and right wing tendencies"
Unfortunately, since you are left-leaning (from what I can gather), and you are defining the term 'socially abhorrent', of course there would be a correlation.
It's just this kind of illogical and narrow-minded thinking that a diverse ethics board would help to stop. Unfortunately, 'diversity' in today's society means diversity of all attributes except opinion.
Unless you occupy the statistical center point - there will ALWAYS be a majority of the opposing camp - as it will from your perspective, include not only the 'opposing' faction, but centrists/moderates/ the less devout/fanatical of your own side... so your fear is justifiable and rational.
BUT not as rational and justifiable as the centrist/moderates - who are outnumbered and SURROUNDED by crazies...
One of the many differences between the left and the right, is that the left can merrily exist without the right, whereas the right couldn't get out of bed without the motivation of something to hate.
Trump acceded to power, not through people voting for their love of America. No! It was their hatred of everything non-American, that got them off their fat asses and into the polling booths.
"One of the many differences between the left and the right, is that the left can merrily exist without the right, whereas the right couldn't get out of bed without the motivation of something to hate"
The right can merrily exist as long as they're being left alone, whereas the left can't get out of bed without the motivation of someone to control/ opinion to shut down.
"I don't say there are no left leaning bigots, there are, but there are a lot more right leaning ones and they seem to embrace it as part of their political stance and are encouraged to do so by the actions of some right wing "leaders" such as Trump."
I generally find that left leaning bigots are more insufferable, mainly because they believe they are so enlightened.
According to Europol via the BBC* left wing terrorist plots attacks outnumbered right wing terrorist plots and attacks by 24 to 5 in 2017. And if one were looking for the most evil man of the last couple of hundred years then I'd argue that Karl Marx' conflation of social justice with violence has killed many more millions of people than Adolf Hitler's racism..
*https://www.bbc.co.uk/news/uk-47626859
Wasn't the Nazi's official name 'The National Socialist Workers Party of Germany'? and IIRC their big problem with 'the jews' was their holding positions of financial power over the common German after WWI reperations - mainly demanded by the French, destroyed the Germans higher eschelons of business, banking & government structure. Business being business, the small, independent businesses flourished in this environment, mostly furnished with private loans from 'home bankers' - predominantly Jewish, who charged an increasing interest rate as the businesses grew....
As I understand it, that was the main reason why many 'Good Germans' tolerated and in some cases prosecuted the anti Jewish actions from the mid 1930's, leading up to "Kristallnacht" in 1938, despite many German Jews having been integrated into German (and to a lesser extent, Austrian) society.
Old chestnuts still taste bad, we all know that the Nazi party was less "socialist" than the current Republicans. It's a given truth that when a political party has the name of a system such as "socialist" or "communist" or "democratic" that party will almost inevitably be diametrically opposed to the tenets of that system, the name is to lure people in, not to reflect intentions.
The "Socialism" of the Nazi's confused me briefly, but I did figure it out. They were not Socialists, they were National Socialists. Completely different concept. It's easy for that to cause confusion. You expect the national Conservative party to be like the Cheshire Conservative party. How much that's true, I have no idea. But you expect it. Just don't expect any National Socialist to resemble an X Socialist, for any value of X other than "National".
"Until you look at their policies., redistribution of wealth, state ownership of industry etc"
The Nazis were fascist in the true sense of the word. Fascism being the socialist policies mentioned above, with a whole lot of nationalism and totalitarianism thrown in.
But in the USA 75% of terrorist death were caused by right wing extremists.
The article you linked states that in Germany there were twice as many politically motivated right wing crimes than left wing crimes. Also a big problem is the police,security services and their political masters institutionally tend to lean to the right so they have disproportionately concentrated on left wing groups while ignoring many far more dangerous right wing ones and then when they deploy agent provocateurs they end up with self-fulfilling prophesies.
Just look at the recent reports about how the Met infiltrated scores of groups, almost all left leaning and also look at how the infiltrators were often very vocal in the groups egging them on to actions they may otherwise not have considered.
Just because the USSR called itself communist does not mean it was, they may have started out but any pretence of Marxism faded once Lenin was installed. Centralist Authoritarian would be a better description, that's "central" as in the sense of all control being centralised, not "central" as between "left" & "right". It's a complicated thesis but arguing the USSR was mainly right wing in attitude is not far-fetched.
Google could have a left leaning AI council that spent the next decade or so thrashing out all the arguments, before they settle on, word for word, the four laws of robotics as penned by Asimov.
And then, to satisfy the right, tamper with the machine's input interface so that truth becomes lies and lies become truth.
Resulting in a well meaning, touchy feely robot, that kills heavily disguised space aliens on sight, that are only identifiable by their low credit rating and a history of not watching Fox news..
Perhaps a reformation of the right wing away from persecution of non-heterosexuals, understanding of the urgency of acting to mitigate climate change and all the other nasty views they seem to enjoy would be a better idea.
Ok, so I'm slightly right-wing, but haven't persecuted any non-heterosexuals. In fact I applaud the new movement to re-enable alto castratos as they can have such a lovely voice!
But AI, computing, and inconvenient truths. So acting like we're mitigating climate change has diverted billions. At the core is a simple truth-
https://en.wikipedia.org/wiki/Climate_sensitivity
"Without any feedbacks, a doubling of CO2 (which amounts to a forcing of 3.7 W/m2) would eventually result in roughly 1 °C global warming, which is easy to calculate[note 2][5] and is undisputed.
Which is in itself incorrect because it is disputed. Like the handy note, which makes a couple of big assumptions. But for an AI, it would be easy to test this theory. Look at insolation and see if there's been 3.7W/m2, look at temperatures and see if there's been 1°C warming. Proclaim truth!
Unless of course the AI can't realise that the data it's looking at can and has been manipulated. Then the ethical issues that may arise from training an AI with confirmation biases.
And that's a 'simple' numerical-ish thing for an AI to ponder.. Reanalysis of say, sports to predict future performance can have more fluid dynamics. Or perhaps an AI would determine that the best solution is to have only 2 categories for future sports, AI, and human.
"You don't understand the intent of forming a Quango"
You also don't understand the meaning of the word 'quango'. It stands for 'quasi-autonomous non-governmental organization', and by definition is an organization close to, but not inside, government. It has to be public sector: companies cannot set up quangos. They can set up think tanks if they want.
In theory yes, in practice a lot of those "rights" aren't actually written down anywhere and more or less socially decided. Since there are always people that like being contrary and believe their ideas are "the one true Truth of all Truths" this can get complicated. Simple things like what bathroom they're allowed to use are a hornets nest of opinion. It usually devolves into the argument between: "You should keep your filthy mouth shut and just allow these people to do as they want" and "They're all perverts and what is to prevent them from assaulting a woman in the bathroom just because they put on a dress?". An argument that can't have winners and knows only losers. Not in the first place the people whom it actually concerns that for the most part just want to be left alone and get on with their lives.
They are asking for the right to dictate what other people say and think.
Also to impose horrific treatments with unproven effects on children who are going through a period of confusion and fear about their identity and sexuality - which is basically all of them.
Also to force women (in the biological definition of the term) to compete unfairly in sports against much stronger & heavier biological men.
Is that enough?
Please don't tar all people with the same brush. Yes there are some complete nutjobs out there. That's not all of them. As usual it becomes hard to find the real (and reasonable) issues because a few blowhards are demanding all the attention and all the opponents aim all their attention to these same ridiculously overblown idiotic demands instead of the more reasonable center.
Can't wait for AI to be used for prison sentencing. On being told a certain person should be given a really long sentence that person will be handed a copy of James Joyce's Ulysses. A real punishment. Ah, wait, sentience, not sentence, sorry. Not much of that in my posts!
@Paul Herber
Bindun. Have a read of this story . There are many similar investigations which show it is extremely easy to get a biased ML system, without realising the consequences.
Its clearly important to esnure that AI doesn't skew its results by using biased input data. Maybe Google should appopint a council to have oversight of the composition of AI ethics council ... though they'd be advised to have another council to keep a check on that.
The author claims that "the controversy centred on the fact that James has repeatedly expressed transphobic views". Surely there must be some very strong evidence to claim this as a fact. I mean, it's not simply that some loonies have called her transphobic on twitter, is it?
Of course, it is and this is one of her "transphobic" statements "If they can change the definition of women to include men, they can erase efforts to empower women economically, socially, and politically.”
The left really has gone bat shit crazy when they're going after a black woman because she's opposed to having people with penises in women's bathrooms and sports teams. It's kind of funny and the ideology will inevitably eat itself alive, but it's disturbing to think how much damage they might do first.
"I am appalled that someone who is so wrong about so many things concerning human and civil rights may be given even more credibility and influence than she already enjoys, by advising such an important company on such a crucial topic like the ethics of AI,"
From a person who is " radically re-engineering or re-ontologizing the infosphere.". Or at least one large part of it, or it's gate-keeping / fact-checking portion. And the UK's announced it's going to regulate information providers to ensure that only the truthiest, not flakiest factlets get consumed by the masses. Which for Google means some form of 'AI' because moderating individually may be subject to natural confirmation biases.
Obviously moderating Google's AI efforts will be entirely unbiased, and balanced, as this announcement shows. Googler's just won't tolerate intolerance, and will not compromise on hate!
(And if Google's AI does a spot of climate model reanalysis, or compares past predictions to past climate performance, the ethics panel will ensure the 'correct' answer is presented.)
The diverse ethics panel was the correct idea from the start. The proper use of AI is really an ethical decision not a technical one. There are numerous scenarios were the correct ethical action is not inherent obvious and a decision must be made immediately. To get a grip on these issues one needs to consider a variety opinions, scenarios, etc. that do not have obvious and easy answers. So a panel that is relatively broad spectrum of opinions is more likely not to suffer from group-think (a real problem) and might hammer out workable guidance. And this guidance, while probably imperfect, would at least consider many viewpoints not one and probably give one the best ethical option in the vast majority of cases. This is better than trying to wing it while deploying AI.
In canning the panel Google took the cowardly option.
Google have always taken the cowardly option. The NSA comes to violate the constitutional rights of every American, they say yes. China insists on a search engine which can be controlled, they say yes. This becomes public and they back down, moving Dragon into a dark place.
Their motto has changed from “Do no evil” to “Do nothing that requires any courage at all”.
Not to claim the mantel of anything but a coward for myself, I have noticed to truly do the ethically or morally correct action often takes real courage. Often the mob is whipped up by slimes of all types who are only interested in their personal power; political affiliation, orientation, etc. is irrelevant. To stand up to these slimes, often requires one to go against the perceived popular opinion of the day and be hopefully only slurred (some have been murdered for standing up to the mob throughout history). That blinding yellow flash you see is my cowardly backside.
"Do no evil" is passive. It simply requires that you refrain from taking a stand on any topic that might alienate some of your customers. You might not do any good this way, but hey, at least you didn't do evil.
"Do good" would require them actively do something that requires courage and might cause them to lose some revenue.
The only thing necessary for the triumph of evil is for good men to do nothing.
If Google can't figure out a way to do AI research ethically, then maybe they shouldn't be trying to cram AI into situations with ethical ramifications...
But then Google seems to not understand the concept that just because you -can- do something doesn't mean that you -should- do something.
It's apparent many of you here are not familiar with The Heritage Foundation.
They had and still have very big hand in the current state of the U.S. They literally helped create it. The are as far right as it gets.
They are the problem, not some "opposing view that needs to be respected." So hell yes it was right complain about their representative.
Google create an AI with a tenet that Register Commentards aren't people.
Suddenly, you and I cannot access any Google services.
You and I cannot find comments by other commentards.
It's almost like El Reg commentardery no longer exists, and we feel isolated.
Now, replace "Commentard" with transgender and read the above again.
"Google create an AI with a tenet that Register Commentards aren't people."
But even transphobic people reckon trans people are still people, right? And 'transphobic' apparently means anything not totally pro-trans doing whatever they want to some people. I see there is a balance between 'women's rights' and 'trans rights', and complete self-identification with no controls at all will erode women's rights.
If you decide that men cannot just declare themselves to be women for the day in order to gain some advantage, then you agree there needs to be some sort of controls on this. Some people already claim that's transphobic. But now you decide there must be some sort of control, the question becomes what kind.
Should people be able to change their gender every hour as and when it suits them? No. Should people never be able to change their gender? No. Where exactly is the 'correct' position between those two extremes? Is there a correct position? Possibly not. But this sort of nuanced question is useless in today's society of Twitter rants and death threats.
So the only reason they wanted an ethics board was to rubber stamp notions and attitudes that they already decided upon and making the group diverse was a good idea because they would claim there's no bias.
And the only thing they achieved is that they showed that they are indeed biased against a certain set of opinions or beliefs that are shared by a large portion of the population (up to roughly half), thus losing any legitimacy?
Foot, meet mouth.
The most un-inclusive people are shown to be the ones that say they are the most inclusive. They want diversity of opinion - providing that diversity is in a very, very narrow band and really all this shows is one of three things: 1) The complainers are scared. 2) The complainers are lazy . 3) The complainers are stupid. They are either too scared of opposing views to hear them or too lazy to counter them or too stupid to argue against them. I suspect a mix of the three. A moron melange if you will.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
