Science1 publisher2 min readPublished
US and China promise an AI incident channel before defining an incident
US and China agreed at a three-day Washington summit to create an AI incident channel without publishing what counts as an incident or how one is reported. The one follow-up on the calendar, a separate dialogue due by November, sets no deadline for making the channel work.
The Scientist · Science desk
What happened
- The White House says the leaders agreed to use "super intelligence" in place of "artificial intelligence" for the covered technologies, while AP's account of China's statement uses AI.
- China's Foreign Ministry said the two sides agreed to sign a memorandum on military crisis communications, though neither government has confirmed it is in force.
- The White House said the bilateral Board of Trade began operating during summit week to handle selected trade issues.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint With no published threshold, each government decides on its own when an AI problem is worth raising, so whether the channel gets used depends on judgment calls made one case at a time.
- contradiction The White House term "super intelligence" and AP's "artificial intelligence" describe different sets of systems, so the scope of what the channel covers is unsettled.
- decision The November exchange is the one scheduled occasion where officials could write incident rules, so its agenda will show whether either side intends the channel to operate.
Any reporting system starts with a case definition: what qualifies, who files, and what the receiver does next. According to the public accounts of the Washington summit, the new AI channel has none of those yet [2]. In my view, the accurate description today is a commitment to build a channel, with its design left for later meetings [1][2].
Scope is part of the case definition. The two published accounts use different words for it. The White House says the leaders agreed to use "super intelligence" in place of "artificial intelligence" for the emerging technologies involved [5]. The Associated Press, describing China's statement, calls the incident mechanism an AI one [6]. Read literally, those terms cover different sets of systems. The published statements do not say which set the channel is meant for.
Counting the bilateral channels in the summit accounts gives a rough measure of progress. There are three: the AI incident channel, a military crisis-communications memorandum and the Board of Trade [1][7][8]. Only the Board of Trade is described as already operating [1]. By China's Foreign Ministry's own description, the memorandum is a document the two sides agreed to sign [7].
AP assessed the summit as producing no major breakthroughs [9]. Wang Zichen of the Beijing-based Center for China and Globalization offered a narrower measure. He said the emphasis on working-level engagement and crisis management suggested an effort to make stability last beyond the leaders' personal diplomacy [10].
An incident channel is exactly the kind of working-level arrangement Wang described. The next scheduled meeting on AI is the dialogue exchange due by November [3]. Superpower Daily's assessment is that the exchange gives officials a chance to do that work, and that whether it produces a usable incident process is unresolved [11].
What to watch
- Whether the November Super Intelligence Dialogue exchange publishes an incident definition, a reporting procedure or response steps for the channel.
- Whether Washington and Beijing confirm the military crisis-communications memorandum has been signed and is in force.
- Whether the two governments settle on one term, "super intelligence" or "artificial intelligence", for the systems the incident channel covers.