ChinAI #204: Ant gets Antsy about Trustworthy AI
How Ant Group leverages AI to defend cybersecurity
Greetings from a world where…
the Hawkeyes keep finding ways to win with the nation’s worst offense
…As always, the searchable archive of all past issues is here. 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).
Feature Translation: The Battle between Internet Giants and Cyberattacks
Context: This past September, around the time of the Mid-Autumn Festival, researchers at the University of Science and Technology of China ran a little experiment on their colleagues and students. They distributed phishing emails that advertised “free moon cakes” to 40,000 faculty and students. More than 3,100 students and 400 faculty failed this test. As China’s AI ecosystem matures, addressing security risks will become more important. According to this week’s xinzhiyuan [新智元] article (link to original Chinese), “In the past, AI was driven by knowledge and data, but now, the industry has come to a stage where safety/security and controllability are the core.”
Key Takeaways: In recent years, Chinese tech companies have jumped on board the “Trustworthy AI” bandwagon.
JD.com, Tencent, Ant Group have all published guidelines on the topic. Qihoo 360 has invested in key cybersecurity labs (e.g., its 0kee Team and Vulcan Team). This week’s article profiles Ant Group’s Tiansuan Lab, which focuses on trustworthy AI and intelligent risk control. Since Ant Group manages the massive mobile payment platform Alipay, it has to defend against countless hacking attacks and scams every day.
Wang Weiqing, who heads the Tiansuan Lab, bases trustworthy AI on four principles: privacy protection, robustness/stability, interpretability, and fairness. Earlier this year, his lab open-sourced its “Ant Mirror” [蚁鉴] platform for making AI algorithms more secure — which the article described as a “big event” in the field of trustworthy AI.
Fight fire with fire, fight AI with AI?
As cybercriminals increasingly use AI technology to deceive users, Ant Group’s intelligent risk control technology is also using AI to identify fraud.
One example is the use of an AI bot to call users through a “wake-up hotline,” in order to check up on whether the user has been deceived by a scam. Per Ant’s “2022 Anti-fraud Governance Semi-Annual Report,” its AI wake-up hotline has daily interactions with 43,000 people, and the average conversation lasts more than 90 seconds.
Another example is the use of contrastive learning techniques to judge whether there are inconsistencies in a user’s words and speech, so as to identify whether they have been deceived.
First Half of TRANSLATION: In this battle, the Internet giants have fought with them for 18 years
ChinAI Links (Four to Forward)
Must-Apply: 2nd Mini-Conference on Politics of Emerging Technologies
We’re running it back — big ups to Baobao Zhang for leading this initiative!
Following the success of our inaugural Politics of Emerging Technologies Mini-Conference, we are excited to organize the 2023 Mini-Conference. We invite panels and papers at the intersection of politics, policy, and emerging technologies…Some examples of research topics include (but are not limited to) digital surveillance in public and private spheres; cybersecurity and international relations; the ethics and governance of artificial intelligence; automation and the future of work; the governance and regulation of social media platforms; military and civilian uses of emerging technologies; privacy law and policy; biosecurity and the prevention of future pandemics; and digital media and the spread of misinformation. This year we especially welcome submissions related to the theme of the 2023 American Political Science Association Annual Meeting: “Rights and Responsibilities in an Age of Mis- and Disinformation.” See more information here: https://connect.apsanet.org/apsa2023/
Should-read: Comments section on last week’s health codes article
*Shout-out to Gabriele de Seta for adding in translations of Chinese netizens’ comments on last week’s feature article on government overreach and health codes. Some insightful takes in this section.
Should-read: To enter the digital world, industry must first cross the hurdle of security (in Chinese)
Ant Group’s PR team must have been working in overdrive this week. Another article about the importance of digital security for digitalization, with a lot of examples from Ant’s work. Same team that put out the longform article on CUHK that we translated earlier this month.
Want-to-read: Hua Hsu’s new book Stay True
For a sense of what the book’s about, see this review in The New York Times by Jennifer Szalai.
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 an Assistant Professor of Political Science at George Washington University.
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