Your submission is now in Draft mode.

Once it's ready, please submit your draft for review by our team of Community Moderators. Thank you!

Submit Essay

Once you submit your essay, you can no longer edit it.

Pending

This content now needs to be approved by community moderators.

Submitted

This essay was submitted and is waiting for review.

How impressive will the January 24 Starcraft II AI reveal be?

Question

This question is related to the superhuman Starcraft question on the main Metaculus site, the $10.000 Starcraft question, and [1], [2] and [3] Dota 2 questions.


This is a rare opportunity for a very short-term AI milestone prediction: Blizzard & DeepMind are announcing tomorrow (at 10:00 AM Pacific Time) their latest progress on Starcraft II AI [1] [2].

How (un)impressive will the results be?

It's hard to make this question more precise since we don't know how they'll be measuring their own progress. However, here is an attempt:

Will the AI they showcase January 24 be credibly claimed to be capable of beating master-level human players at least once, in a non-trivially restricted version of Starcraft II?

A few months ago Tencent announced the TStarBots system which they estimate have the ability to beat platinum and diamond-level humans (based on informal discussions on Starcraft II forums), and which demonstrated some success against previous bots and humans (see the arXiv paper and the discussion on Hacker News).

We will take a non-trivial restriction to be one not more restrictive than that Tencent system, which only played a specific map, Zerg vs. Zerg, and 1 vs. 1.

In case the announcement concerns another upcoming benchmark, an AI playing a more restricted version of Starcraft, an AI of uncertain capacity, and so forth, the question will resolve ambiguously.

Make a Prediction

Prediction

Note: this question resolved before its original close time. All of your predictions came after the resolution, so you did not gain (or lose) any points for it.

Note: this question resolved before its original close time. You earned points up until the question resolution, but not afterwards.

Current points depend on your prediction, the community's prediction, and the result. Your total earned points are averaged over the lifetime of the question, so predict early to get as many points as possible! See the FAQ.