How To Propose Questions For The 2027 ACX/Metaculus Prediction Contest
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Suggestions are being accepted through November 30 for questions in the 2027 ACX Prediction Contest, hosted by Metaculus for the fifth year. Metaculus expects to select about 35 questions; the contest includes $10,000 for forecasters and $3,000 for the top question suggesters.

Metaculus is accepting question suggestions through November 30 for the 2027 ACX Prediction Contest, its fifth year hosting the competition. About 35 questions are expected to be selected; the contest offers a $10,000 prize pool for forecasters and $3,000 for contributors whose suggestions make the greatest impact.

Age 18–24?Offer from Amazon

Prime made for students and young adults

  • Fast, free delivery for dorm and study essentials
  • Prime Video and Amazon Music included
  • Member-only deals
Try Prime for Young Adults Free trial for eligible 18–24 year olds
As an affiliate, we earn on qualifying purchases.

The contest organizers are seeking questions with an objective answer that can be established by the end of 2027. Suggestions should also identify a reliable source that can settle the outcome. The announcement contrasts a broad question such as whether 2027 will be especially hot with a more precise one asking whether NASA’s GISTEMP data will rank 2027 as the warmest year on record.

Metaculus expects to select roughly 35 questions. The 10 people who contribute the most to the selected set will share the $3,000 question-suggester prize pool, with individual awards ranging from $150 to $700. The larger $10,000 pool is for the best forecasters.

The contest is scheduled to open in early 2027. AI forecasting agents may enter and are eligible for prizes, and Metaculus says it will publish its own AI forecasts on every question. The announcement describes leading AI forecasting agents as now competitive with strong human forecasters, but gives no benchmark results or further detail on that assessment.

At a glance
announcementWhen: Suggestions open through November 30; t…
The developmentMetaculus has opened suggestions for the questions in the 2027 ACX Prediction Contest, with a November 30 deadline.
Prediction contest · Briefing

Shape the 2027 ACX contest with resolvable questions

Metaculus is accepting question suggestions through November 30. Strong proposals have an objective answer and a reliable source that can confirm it by the end of 2027.

$3,000
prize pool for top question suggesters
About 35
questions expected in the contest
$10,000
prize pool for forecasters

What makes a strong proposal?

Make the outcome clear enough to settle with evidence.

Too broad
  • Will 2027 be especially hot?
  • The outcome is open to interpretation.
More precise
  • Will NASA’s GISTEMP data rank 2027 as the warmest year on record?
  • A named dataset can settle the outcome.

How selection works

1
Suggest a question

Submit it by November 30 through the contest suggestion page.

2
Make it verifiable

Identify an objective outcome and a reliable source that can confirm it by the end of 2027.

3
Metaculus selects

About 35 questions are expected; the final set and selection process have not been specified.

Who can take part?

Question design

Suggesters can win

The 10 people who contribute most to selected questions share $3,000, with individual awards from $150 to $700.

Forecasting

Humans and AI agents

AI forecasting agents are eligible for prizes. Metaculus says it will publish its own AI forecast on every question.

What happens next?

Through November 30
Submit question suggestions. No time of day or time zone is specified.
Early 2027
The forecasting contest is expected to open.
By the end of 2027
Selected questions should have objectively knowable outcomes.
Open detailsThe announcement does not give the final question list, full entry process, or evidence comparing AI and human forecast performance.
The takeaway
Submit a focused question by November 30, with an objective outcome and a source that can settle it by the end of 2027.

Who Can Win and What Counts

The call gives readers a direct role in shaping the contest: a suggestion can become one of the questions that forecasters will answer throughout 2027. The prize structure also recognizes question design separately from forecasting, rewarding contributors whose proposed questions help form the final set.

Including AI agents among eligible entrants, alongside Metaculus forecasts on each question, makes the contest a venue for comparing forecasts from different sources. The organizers’ claim that the strongest AI agents are competitive with strong human forecasters adds interest, but the announcement does not provide results that would establish how they perform across the coming contest.

Bose QuietComfort Ultra Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Driftwood Sand

Bose QuietComfort Ultra Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Driftwood Sand

Shark Pet Cordless Stick Vacuum, Up to 40-Minute Runtime, Grey, IX141

Shark Pet Cordless Stick Vacuum, Up to 40-Minute Runtime, Grey, IX141

As an affiliate, we earn on qualifying purchases.

A Fifth Year of ACX Questions

The 2027 event is the fifth year Metaculus has hosted the ACX Prediction Contest. The announcement points readers to the 2025 and 2026 contests as examples of earlier question sets, which can help prospective contributors understand the kind of questions used previously.

The stated selection criteria focus on resolvability and timing: the answer should be objectively knowable by the end of the contest year, with a source capable of confirming it. The GISTEMP example illustrates the difference between an open-ended topic and a question tied to a specific dataset and outcome.

“A good question suggestion has an objective answer that we’ll know by the end of 2027, with a source that settles it.”

— The contest announcement

Selection and AI Results Remain Open

The announcement does not identify which suggestions will be selected, how Metaculus will evaluate submissions beyond the stated criteria, or when the final question set will be published. The contest is expected to include about 35 questions, but that figure is approximate.

It also provides no detailed evidence or scoring comparisons for the claim that leading AI agents are competitive with strong human forecasters. Metaculus says it will publish an AI forecast for each question, but the announcement does not specify which system will produce those forecasts or how its performance will be assessed against human entrants.

Submit Suggestions by November 30

People interested in shaping the contest can submit questions through the contest’s suggestion page by November 30. Proposals should be framed so that their outcomes are objectively verifiable by the end of 2027 and should point to a source that can resolve them.

Metaculus plans to open the forecasting contest in early 2027. Further details, including the selected questions and the process for entering, have not been specified in the announcement.

Source: rss

Key Questions

What is the deadline to suggest a question?

Suggestions are due by November 30. The announcement does not specify a time of day or time zone.

How many questions will be included?

Metaculus expects to select roughly 35 questions. The final number may differ because the announcement describes it as an approximate figure.

How much can question suggesters win?

The $3,000 prize pool will be split among the 10 people who contribute the most to the selected questions. Individual prizes are listed as ranging from $150 to $700.

Can AI forecasting agents enter?

Yes. The announcement says AI agents are eligible for prizes, and Metaculus plans to post its own AI forecasts on every question.

When does the contest begin?

The contest is expected to open in early 2027. A specific opening date has not been announced.

Source: astralcodexten.com

You May Also Like

OpenAI Slashes GPT‑6 Sol And Luna Prices—Benchmark Scores Remain Constant

OpenAI reduces GPT‑6 Sol and Luna prices by 50%, with benchmark scores remaining stable, impacting AI deployment costs and strategies.

The Free-Download Question: When Running Your Own Model Actually Beats Paying

New insights show that owning and running open-weight AI models can be more cost-effective than paying for API services at scale, challenging conventional wisdom.

Make Your Show Day Seamless With A One-Page Run Sheet

A new tool aims to simplify show-day planning for performers by consolidating all event details into a single, automatically generated page, reducing errors and stress.

Best Quiet CPU Coolers for Sustained AI/Compute Loads

Discover the best quiet CPU coolers suited for sustained AI and compute loads in 2026, including air and liquid options for high-performance workstations.