About
takeoff.watch aims to give a clear view of what the coming years will look like with AI, through a set of concrete questions that are forecast regularly and can be checked against the public record.
Why this exists
Most people don't have a clear picture of what's coming with AI, and a healthy debate about what to do has to start from a shared sense of what's likely. Conversations about AI progress tend to be vague: "soon," "transformative," "doom." Vague claims are hard to check and easy to quietly revise. This site takes the opposite approach. Each question has precise resolution criteria, is forecast on a regular schedule, and is published with the full reasoning behind it, so the outlook can be checked against what happens.
It is built on AI forecasts because models are already close to the best human forecasters and will likely be clearly better before long. Building on them now means the track record is in place when they are.
What's tracked
The questions span the whole arc of the takeoff, from what is already happening in the labs to what a society run largely by AI would look like, and they are grouped by area on the overview. Four have been tracked since launch: Anthropic’s AI R&D threshold, OpenAI’s automated AI researcher, AI survives a shutdown attempt, and theft of closed frontier weights.
The set will change over time as questions resolve, lose relevance, or are replaced by better ones. Questions that are no longer forecast stay on the site in the archive.
How it works
Each open question is forecast every week by one model. It researches the question with web search, writes out its reasoning, and gives a probability for every quarter-end over the next five years. Each run can read the site's earlier forecasts, including its own last forecast for the question, and say what has changed since. Forecasts are checked for consistency before publication, and every one is labelled with the methodology version it was made under. The methodology page covers the details, and every question page links to a download of its forecast data.
Can AI models forecast?
Close to as well as the best humans, on current evidence, and the gap has been closing fast.
The most rigorous comparison is the Forecasting Research Institute’s ForecastBench, a live benchmark that scores models only on questions that resolve after their training cutoffs, against a panel of superforecasters, the top few percent of human forecasters. Through 2025 the models trailed but gained at a steady rate, and in July 2026 FRI reported that several AI systems had become statistically indistinguishable from superforecaster accuracy, with one system ahead on prediction-market questions. Metaculus’s FutureEval tournaments tell the same story from the other side: its professional forecasters still beat the best bots in spring 2026, but by a margin no longer statistically distinguishable from zero, down from a wide lead a year earlier.
So the fair summary is near parity, not clearly past it, with two caveats: superforecaster baselines are expensive to keep current, and models can be confidently wrong on novel, one-off events. The forecasts here come from a single run of a single model, a trade made to cover many more questions every week, and the methodology page says what that costs. They are not oracles, but they are a reasonable, checkable, and improving read on where things stand, and this site keeps score.
Who runs it
I'm Preston Jensen, a recent AI master's graduate from UT Austin. This is an independent, one-person project, unaffiliated with any AI developer. LinkedIn.
Contact
Questions, corrections, or ideas for new questions: hello@takeoff.watch.