A Chinese Finance Giant That’s Secretly an AI Company (2017)
technologyreview.com
technologyreview.com
And they have (drum roll) an equivalent of the FICO score! It uses "advanced machine learning algorithms"!
There's really nothing related to privacy or societal implications, it's an advertorial. I don't understand why it went past the radar of MIT Technology Review.
A bigger question from me, is China really that advanced? I hold a neutral position on this, but am now seeing a consistent pattern of articles hyping Chinese cookie-cutter tech (or stuff which is bigger) sounding like Cold War era Soviet propaganda. If they really had some kind of grandiose tech achievements, why bother with this?
In case you didn't realize, the "MIT" technology review has nothing to do with MIT. It used to be basically the alumni magazine (if you're an alum you get it for free with an alum section bound into the back -- does anyone actually pay for the magazine?). A few years back a company took the magazine over and added "MIT" to the front. They basically want to be Wired, so advertorial content is prime content.
I'm laser-focused on the developing markets (although China is already past that stage). Moreover, I hold investments in mutual funds for emerging markets, including those with stock in Tencent (for instance).
But I can't name anything in China except:
* bigger supercomputers
* bigger bitcoin mining facilities
* clone cryptocurrencies
* bigger number of patents filed by Huawei, Alibaba, Tencent, and Baidu
* hiring of Andrew Ng and then him quitting Baidu
* proliferation of cashless payments
* excellent team work of techies and business people resulting in WeChat, which would not have succeeded elsewhere
* I no longer buy the claims of ML applications (speech recognition, face recognition) surpassing human accuracy - I've been hearing these for years and then reading refutations, but yes, sure, let's assume a few algorithms were tweaked successfully
Does it all work, yes. It's far beyond the China where I was doing business a decade ago wondering whether the developers I was in contact with have a functioning brain. But is it "more advanced than Silicon Valley" - nowhere close to it, it's well-managed cookie-cutter tech. If anything, the credit goes to the Chinese tech-savvy business people not prone to the American corporate waste syndrome and forced to dive deeper into the tech.
Can you name 3 (three) projects where the tech was not licensed or borrowed elsewhere and is indeed so advanced it did not surface elsewhere? I'd be sincerely interested to learn of such.
I haven't heard of anything 'pure tech', but there's a big network effect to having the world's electronics industry on your doorstep. If you're building a microscope, you don't have to wait for months for the parts to arrive. If you're building a new kind of centrifuge, you have a massive pool of great engineers who can do it for you, just a phonecall away.
Look at Ant Financial's credit score system known as the sesame score, it is just a huge disaster and embarrassment to the company. Last year, Ant Financial openly promoted the idea to let high sesame score users to exclusively date young college girls, you will only be allowed to contact/interact with those college girls if your sesame score is higher than 750. You have to question the overall integrity of the company and the team behind such ideas.
At the beginning of 2018, users received notification from Ant Financial for viewing their 2017 yearly bill. Ant Financial forced all users who want to view that yearly bill to disclose their sesame score potentially to the rest part of the Alibaba group.
No one from the top management team stepped down after the above two incidents, even when both of them hit national headline. Why? Because uncle Jack Ma's Ant Financial doesn't give a crap on such "tiny" incidents.
With such track record, AI or no AI, I wouldn't trust such a company for handling my personal data and finance. Luckily, there are other far better & more innovative Chinese competitors kicking Ant Financial's ass right now, e.g. Tencent.
> With such track record, AI or no AI, I wouldn't trust such a company for handling my personal data and finance.
China is not America, they value societal stability over personal freedom. It's absolutely fine that you wouldn't trust this institution, but it's not made for you.
And I'm pretty sure the grandparent is well aware of the political situation in China, just take a look at their comment history.
Right, which is my point. They won't mind something like limiting access to college women to those with good scores like America would, so they won't take their money elsewhere. Their different culture causes companies to behave differently than they would in the US.
Some people definitely did mind, since the social dynamics caused lots "pretty women" posting selfies (not all of themselves) to ask those men with scores > 750 for money and/or attention. Someone punned the Chinese name for Alipay (支付宝, zhifubao) into "pay the pimp" (支付鸨, pronounced identically), which might give you an idea how someone who culturally values social stability might come to reject such a company. As far as I can tell, the feature was then scrapped pretty quickly.
the message is clear here - you are encouraged to violate laws to try maximizing company revenue but you must stick to company policies.
it is a social risk to allow companies like this to master AI techniques.
If they achieve to recognize the car in a given picture, and maybe even the type of car, even though it's damaged, that will be really great. With that they may be better than the current best solution I've seen (FB's object recognition, which may not even be on as high a level as they present to us in a blog post). But estimating the damage, means recognizing something that is not a pattern. I doubt even classifying this into three categories like `~100bucks`, `~2000bucks`, `buynewone` would already be hard. I doubt that anybody could do it.
And here's the point: Yes, Chinese IT companies are picking up speed. And I wish them all the success. But they have yet to show something that is really innovated by them. This idea is great, but they need to prove that they can do it before I will clap for their achievements.
There are a few companies already doing similar, https://www.galaxy.ai as an example there are a few others.
From what I can tell, they are producing good guesstimates but not yet human level.
He's a skilled techie with a developed bullshit radar. That already puts him above an average layperson.
Over the last 20 years, there's been untold amount of hype from different sources, each wave of which wanted the public to believe that this time it's different.
Well one fact I can tell about is the expo in my university before I came to USA, the AI technology shows during the expo suprised me, although the mechanism behind problem simple, well in China senario, the nlp, characters recognition is more difficult due to it's culture.
I happened to check out it's individual loans oneline, based its credit system, everyone has a credit number, I can lend my money to thers, it's like customer to customer, it's more like the blockchain idea (decentralize), I like it, we don't need bank or anything other agents to operate for us, although the credit number can be compromied by Ant, the idea is real, and used by Chinese people, it's happening, it's a start.
I know here is a different country, different system, basically payment dominated by cash and credit card, which I was really not used to, no offense, living here I feel more traditional regarding payment and shopping.
Of course there are so many things good and better than China, but China is so big, there are technologies advanced indeed. Back to Ant, there are certain ways they promote their products I don't like, the truth is they do innovate Chinese shopping, loans and financial payments, and mostly it's working well, it's more align to the society develop.
Is this legal in the U.S. and E.U.? These AI systems can increase profits by saying who should get loans, but not why they should. If a U.S. business can't give clear criteria on how it decides on whether or not to grant loans, and instead allows an AI to decide based on any possible unknown criteria which could include ethnic profiling, then wouldn't they get sued soon enough?
-It's hard to find out why you were denied a loan and full transparency is not required. You could ask for how the company has processed your personal details and try to find out but hard to know and hence hard to complain. Even GDPR doesn't change that a black box model for the modeller is incomprehensible to a customer.
-Companies can to a certain degree operate as they want to. (I think that's fine too. Business should benefit both parties.) Discrimination is not always very clearcut and often on proper grounds (think risk-based).
-In the Netherlands companies cannot automate denial of service / business. There has to be a human element to that.
-Actuaries have long learned not to use for example names in modeling. That would be a good example of adding better risk insights through illegal discrimination (names give a good delineation of ethnical background, at least a for few generations).