ChinAI #165: AI = Artificial Challenged Intelligence
Plus, SenseTime's AI Ethics Committee violates its own rules?
Greetings from a world where…
we all drag our favorite newsletters from the Gmail promotions tab into the primary inbox, right?
…Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors). As always, the searchable archive of all past issues is here.
Update: SenseTime AI Ethics Committee
In a previous ChinAI issue (#164), I criticized SenseTime for not naming the members of its much-hyped AI Ethics Council in its recent sustainable development report (published in Chinese). Someone affiliated with SenseTime, who wishes to remain anonymous, reached out to me and argued “it’s not a mystery who the ethics committee members are.” He said the members are listed on the last page of the sustainable development report.
Here’s where it gets confusing. The last page gives two lists of names: one list of six co-authors and another list of eight steering committee (指导委员会) members. It’s not clear which group matches up with SenseTime AI ethics council.
If it’s the six co-authors, then the council consists of:
James Ong, an entrepreneur who runs the AI International Institute, an organization that lists 5 main achievements since its creation in 2017. None of them are directly related to AI ethics and governance. One of them is being “invited to participate as FRIENDS of SLINGSHOT 2020” — a Singapore startup competition.
Bao-liang Lu, a Professor at Shanghai Jiao Tong University that researches how to make machines better understand human emotions. He is working on a project to investigate age, gender, and cultural differences in emotion recognition.
The other four are SenseTime employees.
However, if the AI ethics council actually maps onto the eight steering committee members listed on the last page, then the makeup of SenseTime’s ethics committee violates one of the few committee by-laws. In fact, my anonymous source claims that the steering committee is the AI ethics council, even though he agrees it’s not super clear in the report.
Crucially, on page 40 of the sustainable development report, SenseTime states that “it is mandatory that external members should not make up less than 1/3 of the seats, so as to absorb different backgrounds, perspectives…”
Well, the eight-steering committee consists of six SenseTime employees and just two external experts: Xue Lan, a Dean at Tsinghua University, and Weidong Ji, a Dean at Shanghai Jiao Tong University.
2/8 is less than 1/3, which means the external members hold less than 1/3 of the seats on the council. Somehow, if the steering committee = the AI ethics council, SenseTime violated one of its own core bylaws when creating its ethics council.
To put it gently, this significantly damages any hopes that this AI ethics council would bring in experts from diverse fields to vigilantly provide transparency and guidance on SenseTime’s development of AI.
Feature Translation: Comics on Artificial Challenged Intelligence [人工智障]
One trend I’ve been following is the use of the term 人工智障 to make fun of the limitations in AI (see ChinAI #144). This week we look at some comic strips on the subject, produced by 黄一刀 on December 3rd, which seems to be pretty popular (100k+ reads in two days and a lot of comments)
The first comic strip makes fun of sweeping robots (note: each bullet point gives the translated text for one panel):
If you think of the so-called smart sweeping robot as a maid, you will find that it . . .
Eh, can’t sweep this . . . too narrow
Can’t get over there, there’s something blocking it
I can only sweep half of this room, unless you take away all these things
After getting it, I’m much more diligent about cleaning up the house, because I need to clean up all the obstacles for it.
Ai, ai, ai ~ not bad, not bad ~I can sweep it
One of the comics captures the frustration of talking with a customer service bot:
Bot: “Hello, I am a smart chatbot ~ you can directly state your question, or type it out…
…for example, if you want to inquire about ‘after-sales service,’ just directly say: ‘after-sales service’ ~~”
Person: “Wtf?? It’s this smart now?”
Person: “Order tracking”
Bot: “…. ….. Your voice is naturally nuisance-free ~
That spring, she also asked me this question…”
Bot: “Hello, I am a smart chatbot ~ please…”
Person: “I don’t know what I was expecting.”
ChinAI Links (Four to Forward)
Must-read: Online public discourse on artificial intelligence and ethics in China: context, content, and implications
Yishu Mao and Kristin Shi-Kupfer’s article in AI & Society finds that Chinese discussions about AI ethics are “extremely lively and diverse.” They collected 1173 WeChat posts and 1123 Zhihu threads on the subject of AI ethics. What was cool to see is that their breakdown of key authors on AI ethics on WeChat matched up with a lot of the accounts ChinAI translates:
The public accounts of media outlets, especially those focusing on the tech industry such as Leiphone.com and Jiqizhixin.com, published a further 18.2% of the articles. Individuals and organizations in the tech industry accounted for 18% of the articles. Among them, Tencent Research Institute, the social research arm of the Chinese internet giant, is the most prolific account among all types of authors. Since 2017 it has published numerous articles about potential ethical issues associated with AI, introducing global and especially European ethics principles and regulations aimed at the Chinese public.
Should-read: China’s Queer Internet is being Erased
For Rest of World, Lavender Au and Weiqi Liu write about how queer Chinese people struggle to find freedoms online. These points on the dominant gay dating app Blued were striking:
Though these apps present themselves as allies to the gay community, they have aligned with the censors. Blued assigns each user “rainbow credits,” which they deduct if users violate community regulations. Leo has found this includes trying to organize an activity.
When a user loses credits, their profile faces more restrictions, the final stage of which is being frozen. Blued’s parent company is increasingly gathering a monopoly over queer online interactions — in August 2020, it bought the largest lesbian dating app, Lesdo, which it shut down this year.
Should-read: Singapore’s tech-utopia dream is turning into a surveillance state nightmare
Another impressive piece in Rest of World. What an intro sentence by Peter Guest: “On a hard disk somewhere in the surveillance archives of Singapore’s Changi prison is a video of Jolovan Wham, naked, alone, performing Hamlet.”
Should-read: The Fall and Rise of Techno-Globalism: Democracies Should Not Let the Dream of the Open Internet Die
I’m late to this, but I recently caught up on this excellent Foreign Affairs piece by Graham Webster and Justin Sherman. They outline the case for a “renewed and pragmatic embrace of techno-globalism” when it comes to global governance of technology.
Thank you for reading and engaging.
These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is a postdoctoral fellow at Stanford's Center for International Security and Cooperation, sponsored by Stanford's Institute for Human-Centered Artificial Intelligence.
Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99