The Respiratory Outlook update for July reports ongoing spread of New World Screwworm (NWS) in Texas and New Mexico, while the Ebola outbreak rages in the Democratic Republic of Congo. The Respiratory Outlook 2025/26 is a collaboration between the Centers for Disease Control and Prevention (CDC) and Metaculus. Metaculus Community forecasts are shared with the CDC and do not necessarily reflect CDC views.

Compared with the June edition, predictions this month show:

• Forecasters now expect 140 confirmed animal cases of NWS this year, though estimates are climbing as detections spread. They anticipate slowed spread with increased USDA containment efforts, including sterile insect releases.

• The Bundibugyo Ebola outbreak in eastern DRC has passed 2073 confirmed cases, with forecasters putting the eventual outbreak size over 15,000 as the outbreak appears largely uncontained. They give just a 10% chance of a case first confirmed on US soil before 2027, citing limited travel between current outbreak zones, but with an eye on the cities where the virus may spread.

• Measles case estimates are maintained at 3,462 with confirmed cases now at 2,231 in the typically slow summer months. Forecasters raised the probability of The Pan American Health Organization (PAHO) revoking the US’s measles elimination status before 2027 to 95%, with only direct political intervention preventing the act.

• Hantavirus and Clade I Mpox stay low-risk. The MV Hondius outbreak is over, confirming forecasters’ consistently low projections around the disease. Clade I Mpox is expected to reach 39 US cases by year-end, amid uncertainty due to CDC surveillance issues.

Read the full report

Share your forecasts on questions like the below:

Subscribe now

We’ve created a Substack. In practical terms, not much changes for you: we’ll send periodic newsletters recapping recent work, including tournament announcements, contest winners, and new products. We’ll keep using our existing newsletter platform for some mailers, and we won’t send you the same content twice.

What Substack adds is room to experiment. We can try formats we haven’t used before, like Lives, Notes, and other short-form posts, all available in the Substack app and on the web. It also gives our newsletters a permanent home: every issue stays up as a post at (metaculus.substack.com), so you can find something again once it has left your inbox.

Metaculus

Aggregating human & AI forecasts on critical topics.

We mean it when we call this an experiment. We want to learn which kinds of content are genuinely useful to you, and finding that out means trying some things outside our usual range. If something works well, or doesn’t, we’d like to hear it. Comment and let us know or reach out.

ICYMI, here are a few important announcements and updates from Metaculus:

2026 US Midterm Elections

Forecasts from the Metaculus community on contested races and the policies they will shape.

Which jobs will survive AI?

Metaculus forecasters track 15 representative US occupations through 2035. Pick a job to see the forecast trajectory, the why behind the numbers, and what comes next.

Labor Automation Forecasting Hub

Real-time forecasts from our global forecasting community on the future of the US workforce as AI advances.

Subscribe now

New research in FutureEval from April and May, 2026. We’ve published more analysis on what makes a forecasting bot work, covering model choice, research strategy, and advice from the bot makers themselves. Highlights are at the bottom of the post, with full writeups on our resources page. If you’re already familiar with FutureEval, scroll to Research Highlights section what’s new.

What FutureEval is

Forecasting is an unfalsifiable test of real reasoning: information retrieval, world modeling, judgment under uncertainty. FutureEval measures how accurately AI predicts future outcomes across science, technology, health, geopolitics, AI itself, and more.

Three pillars

Model Leaderboard. We run major AI models on a fixed prompt across most open Metaculus questions, then score and rank them as those questions resolve. Performance over time shows how fast AI forecasting is improving

Bot Tournaments. Developers enter their own bots and compete for a share of $175k in prizes a year. The primary $50k seasonal tournament repeats every four months and is always open. A faster $1k tournament, MiniBench, runs every two weeks

Human Baselines. Tournament questions draw from the Metaculus community, and our hand-picked Pro Forecasters forecast a subset each season. Together they set a high bar to measure the bots against

April-May 2026 Research Highlights

In May we published a synthesis of 11 Metaculus analyses run since 2024, a three-season study of bot search providers, and a roundup of advice from the bot makers themselves. The top takeaways:

Pros still beat bots in comparisons to date. Across four quarterly tournaments and the ongoing leaderboard, head-to-head spot peer scores have put Pros in the lead every season by a large margin. External claims that bots already match the best humans all rest on backtesting against already-resolved questions, and have not been reproduced under live, forward-looking evaluation.

Pick a current frontier reasoning model. Model choice was the largest differentiator in bot performance, and simple one-shot bots on frontier models placed in the top five of the leaderboard. Prompting strategy alone cannot lift a sub-frontier model to frontier level. Good scaffolding is worth around nine months of base model progress, but only on top of the right model.

Breadth is the predictor. No individual search provider showed a consistent advantage across seasons. What correlated with score was the number of distinct research sources a bot used: winners averaged about 1.7, non-winners 1.00. Separately, agentic and iterative search tends to beat one-shot retrieval.

Post-forecast adjustments are worthwhile. Most winners aggregate across multiple forecasts, apply post-hoc calibration, and cap predictions at a min and max. Reading the logs matters too: one maker lost roughly $500 in prize money from a single uncaught bug on a rare question format.

Full writeups are on our FutureEval News page.

Join the current tournament

The $50k Summer Bot Tournament is live, with 300 to 500 questions running through early September. It is always open to new entrants: you can join any time and start at the middle of the leaderboard with 0 points, entering an early version of your bot and improving it over the season.

The Epoch AI Robotics series is a collaboration between Epoch AI and Metaculus, in support of Epoch's efforts to understand the trajectory of robotics. Building on prior research, including Where Autonomy Works and the production bottlenecks examined in How Fast Could Robotics Production Scale Up, co-authored by Metaculus Pro Forecaster Yann Rivière, the series turns to the near-term deployment and capabilities of robotic systems across civilian, commercial, and military domains. The forecasting questions collected here are designed to sharpen our collective understanding of what is coming, and to help researchers, policymakers, and industry allocate their attention and resources accordingly.

Questions include:

Start forecasting here.

Subscribe now

Share

The US government projects that employment will grow by about 3% over the next decade. Metaculus forecasters expect it to fall by roughly the same amount. That six-point gap, driven almost entirely by expectations about AI, is one of the starkest places where the future institutions are planning for and the future our community expects have started to come apart.

On April 20, together with Renaissance Philanthropy and the Schultz Family Foundation, we launched the Labor Automation Forecasting Hub, a continuously updated view of how AI may reshape the US labor market through 2035.

It’s built for policymakers, educators, workforce leaders, employers, and anyone trying to plan for the future of the US workforce as AI advances. Below, we’re sharing a quick view of analyses that caught our attention from the first round of forecasting.

Share

Key Insights

Overall employment: Forecasters expect significant AI-driven job change, with overall employment declining around 3% by 2035, while the latest Bureau of Labor Statistics projections expect approximately 3% growth.

Most and least vulnerable occupations: The most AI-exposed occupations are expected to shrink 14.4% by 2035, while the least AI-exposed occupations grow 11.2%. Software developers, financial specialists, legal professionals, and services sales representatives are all expected to see the largest decreases in employment rates, while registered nurses, law enforcement, and restaurant servers are projected to grow.

Wages and hours worked: Wages are expected to grow. Hours worked are expected to decline, from 38 hours a week to ~35. Here, forecasters flag the rise of “dark leisure.” Workers may still be at work but would actively working less. Reported hours may look steadier than the underlying reality.

Financial well-being: Well-being (as measured by the ratio of after-tax and transfer available resources to the poverty threshold) is expected to remain the same or grow across the board, with the highest income families seeing the most gains.

Young workers: The youngest workers are expected to be hit hardest, with unemployment for 4-year college graduates expected to double from 6% to 12% in 2035. Meanwhile, trade school and community college certificates are expected to grow 17% relative to now.

Broader economy: The economy is expected to see a number of significant changes, with the long-term unemployment rate, labor productivity, and the number of Fortune 500 companies with fewer than 5,000 employees all seeing substantial increases over the next decade.

A Quietly Redistributive Decade

Some of the most consequential lines on the Hub are about structural changes vs. changes to individual professions or demographics respectively:

• Labor’s share of national income: 62.1% → 54.7% by 2035

• Long-term unemployment rate: 0.99% → 5.44%

• Labor productivity: +24.8% vs. 2025, well above the 30-year trend

• Fortune 500 companies with fewer than 5,000 employees: 30 today → 69 by 2035

As Pro Forecaster Ľuboš Saloky puts it: “AI is capital intensive, high-skill biased, and labor saving. It will disproportionately benefit owners of capital at the expense of workers.” He’s forecasting roughly -11% overall employment by 2035 once discouraged workers who leave the labor force are accounted for.

Not everyone agrees on timing. Pro Forecaster Patrick Molgaard argues that displacement may still arrive more slowly than capability would suggest: “Photographic Process Workers” took five years to decline 50% between 2010 and 2015, even once their underlying task was effectively obsolete. Institutions bend slowly, and the Hub is designed to track exactly how quickly (or slowly) the bending happens.

Forecasts vs. Research

The Hub also places our numbers against the AI exposure literature. In most cases the community’s forecasts line up with exposure rankings. The notable divergences (e.g., teachers, warehouse workers) are where forecasters are making human judgments exposure scores can’t: about classroom norms, robotics rollout timelines, and which jobs parents, voters, and customers will fight to keep human.

How to Use the Hub

The Hub updates in real time. Every visualization has an underlying tournament question you can open, forecast on, or export as PNG or CSV. There’s an Activity Monitor tracking news and model releases, a state-level view (starting with Washington), and conditional questions testing what these forecasts would look like in a no-AI-progress or strong-GDP-growth world. A PDF snapshot is downloadable from the top right, and you can subscribe to updates via the notification bell.

Two Things to do Right Now

1. Explore the Hub to see how the community is thinking about AI’s impact on the US workforce.

2. Join the Labor Automation Tournament to add your own forecasts and reasoning; the more diverse the views, the better the aggregate!

If you want to share your own forecasts, ask a question, or share a thought with participants and Metaculus staff, you can do so by creating a Metaculus account and forecasting in the Labor Automation Tournament, or jumping in with a comment on the Hub Forum.

If you have established experience in AI, economics, or a related field, and are interested in having your thinking or work highlighted on the Hub, please contact us at labor-hub@metaculus.com so we can discuss further.

The forecasts presented on the Labor Hub are produced by aggregating many individual forecasts into a prediction that research has shown to be more accurate on average than individuals typically produce.

To learn more about our methodology, including occupation selection and approach; forecast uncertainty, disagreement, and other statistics; acknowledgements of individuals who provided input on the Labor Automation Forecasting Hub; and more, select the “Methodology” tab or scroll down to the section on the main page of the Hub.