Will there be another blatant demonstration of AI risks, comparable to Bing Chat, by 2024?
Dec 31

Resolves yes if before January 1, 2025 I think there has been a news event
a) of comparable or higher magnitude than Bing Chat unhinged replies
b) that demonstrates that AI can be really dangerous

I will sell my stock if the market gets >50 traders.

Jan 4: By "magnitude", I meant something like "the amount of concern about AI risks (whether existential or not) expressed in media".

Some methods I'll use to judge this

- Asking myself: "Would I refer to that news event, if a friend asked, whether there has been some actual reported behavior of AI (intentional or unintentional) that indicates at least one of the following
a) AI will soon develop destructive capabilities
b) AI isn't aligned by default (e.g. AI resisting shutdown, manipulating testers, sandbagging)
better than Bing AI responses (currently I find mentioning them helpful for demonstrating we already do have intention / goal-directed behavior and misalignment/unintuitive AI behavior)


- I'll search the top relevant news article on Google News. Then I'll conduct a google search using the site:twitter.com index and see whether I get more results from the Bing story or from a story that comes up in 2024.

- Eyeballing Google Trends. For the Bing reports, I see around +5 % counterfactual search interest corresponding to the week (compared to non-risk AI search interest).
- Eyeballing Wikipedia. For Bing drama, I see around +212% counterfactual views of x-risks (Feb 15-17 was a traffic lowpoint).
However, as you can see, this eyeballing is very subjective, given the large background interest & unclear benchmarks. Therefore, I'll only give them small weight.

Get Ṁ600 play money
Sort by:
predicts YES

I updated the resolution methodology

More related questions