China demands internet companies create governance system for algorithms China's authorities have called for internet companies to create a governance system for their algorithms. A set of guiding opinions on algorithms, issued overnight by nine government agencies, explains that algorithms play a big role in disseminating information online and enabling growth of the digital economy. But the …
Yes, this is sorely needed, worldwide, but not for the reasons stated.
I would welcome an independent overview of all algorithms dealing with social media and finance markets, for example (and probably other areas). The algorithms are hidden behind "trade secret" statements, yet these things control our lives, make decisions about us, from custodial sentencing, through medical systems, financial market manipulation to more banal, yet, probably having the most impact on most people on a daily basis, search algorithms, advertising algorithms, social media post algorithms, biometric recognition (facial recognition).
And those are just the "legal" algorithms, there are probably just as many "dark" algorithms searching around for companies and individuals to attack.
For a look at "what could possibly go wrong", I thoroughly recommend reading Weapons of Math Destruction by Cathy O'Neil.
Algorithms are often "neutral", they aren't necessarily good or bad, but often it is the data they are fed that causes bias. Also, many of the more modern "AI" solutions can't even be tested in traditional ways. AI facial recognition "works", as long as you are a white male, because that is what these systems have mainly been tested on, up until recently. One demonstration of politicians in black America, where many of them were mis-recognised as wanted criminals is an obvious case for an algorithm that should be taken offline without delay, but there are hundreds, if not thousands of similar examples out there, where social or racial bias is unintended, but has huge consequences.
Self-regulation is a good start, but we should be prepared to follow through with independent bodies (also politically independent) that will oversee them going forward, if companies show that they can't or won't put proper controls in place to stop algorithms from producing wildly inaccurate results.
This is fundamentally an essential if we're going to rely on AI and ML for decision making that affects lives, livelihoods or business missions.
It's a pity in a way that it comes out of China, as that may well cause the narrow minded, or worse those with vested interests in lack of transparency, in the 'west' to disregard it on political grounds.
Compare this with the current UK DCMS proposal to repeal Article 22 of the GDPR, which, although it might be less ambiguously expressed, in principle provides for challenge and human review of automated decisions: "The data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her."
Well, the West is recognizing the need for governance of some algorithms. Facial recognition algorithms to which all black people look alike are at least starting to get noticed.
But the thing is that governance of algorithms poses questions. Of intruding on an individual's freedom to do what he likes with his own computer. And his own data, so European-style privacy laws are also problematic.
The problem isn't so much the high value the U.S. political system places on freedom. That's a good thing. The bad thing is the amount of control Big Business has over politicians. (I remember one professor actually having the gall to say that it would be a good thing if politicians were allowed to take bribes, because it would create a "market system" for political decisions, which would clearly lead to more optimum outcomes than everyone's vote counting equally.)
Fix that, and a balance could be struck between the competing issues. In China, there's no need for balance, and so what they're doing is rightly regarded with suspicion.
I started by agreeing with you, then not.
The first paragraph, yes, absolutely, although it is being noticed and calls for banning it are coming up, but those that already rely on such technology, as opposed to human common sense, are often against throwing away their crutch and doing "the right thing".
I would say the EU privacy laws are the one good thing at the moment. It stops the misuse of the data, or having been misused, the individual has the right for the misuse to be put right or punished.
This is also not about individuals, currently, but about large scale AI and ML systems that affect others.
I would also say that the US political system doesn't provide that much value on freedom, at least not individual freedom, it is the freedoms of the highest bidder, which is usually big business. If they were truly about protecting freedom, they would be holding big business in check and not letting it exploit individuals without their permission, instead, the US tends to squash individual rights, when it comes to big corporations and big data.
Also, US politicians have been taking bribes for decades (and it seems to be spreading to other countries over the last 2 - 3 decades as well). What the hell do you think lobbying is? It is buying decisions with the biggest chequebook. Whether that is money directly into the hands of a politician or his campaign fund, or an "if you support this, we'll build offices in your constituency" type bribery.
Given its vast troll army, relentless censorship, rewriting/expunging of history, misinformation campaigns, and the physical reality of the brutal repression/elimination of Uyghurs, it's hard to see why anyone would suspect the CCP of being up to no good with this.
And here we go.
Content deemed dubious in China ? Tienanmen Square.
Content deemed dubious elsewhere ? Good people on both sides, I won Arizona, the election was rigged, etc . . .
Let me give you my definition of dubious content : when it's a bald-faced lie, it's dubious. Otherwise, it's freedom of speech.
You have the right to call me an idiot, that's your opinion and you are entitled to it. You do not have the right to deny historical facts. The Holocaust happened. We landed on the Moon. Tienanmen Square happened. The Twin Towers were not a controlled demolition. The election was not rigged.
Period.
With ML, even the machine doesn't really know hot works and the boundary between the algorithm and the training data is rather fuzzy, so my take on this announcement is that China wants foreign tech giants either to hand over all of their IP or to cease operating in China within 3 years. Maybe both.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
