Independent thought lab · AI risk and the human future
AI may be the most consequential transition our species has faced. CUSP42 Labs measures where AI risk stands, simulates how AI could reshape societies, and asks what it means for humanity, with every assumption on the table.
Two questions
Most attention goes either to the catastrophes or to the productivity. We think the transition has to be studied as one thing: the risks are part of how AI reshapes society, and the reshaping is where most of the risk lives.
The risk question
Loss of control over critical systems. Misuse that lets a few do enormous harm. Accidents in AI-run grids, markets and food systems. AI-accelerated escalation between powers. The slow collapse of shared truth. We track the drivers, and we simulate the odds.
The human question
Work, careers and the ladders into the middle class. Institutions built for slower change. Meaning, agency and attention. What it means to be needed, and what remains distinctively human. We explore it with simulation, structured debate and thought experiments.
What we do
Four ways into the same transition, each one public and free.
Measure
The Civilizational AI Risk Index: 70 anchored questions in six dimensions, from capability to misuse, scored since 2019; daily updates launching soon.
Simulate
What happens to 178 countries when AI arrives: jobs, incomes, trust, institutions and catastrophes, quarter by quarter, with the odds of 28 endings.
Debate
Several AI models argue the same question in rounds, on the record, before any conclusion: for questions where one fluent answer is too fragile.
Imagine
The human question has no dataset yet: work, meaning, agency and what remains distinctively human, explored through structured thought experiments.
What a simulation shows
On today's AI trend, roughly one simulated future in three meets a global AI catastrophe before 2050. Most of the others are not catastrophes but deep, uneven change: broken career ladders, strained public finances, eroded trust, and in some, real shared prosperity. Which world we get depends on choices that are still open.
Scenarios under stated assumptions, not forecasts. Open the model, change one assumption, and see what happens.
How we work
A thought lab earns trust by showing its working. These are the rules we hold ourselves to.
We show what stated mechanisms imply under stated assumptions. Change the assumptions and the future changes: that is the point.
One simulated history is one draw of luck. We report how often outcomes happen across many.
Every index question has its anchors; every simulated change traces to a mechanism and a data source.
One set of rules for every country; no regime labels; no country blamed for a war on our posters.
Our tools structure disagreement and expose assumptions. They do not hand judgment to a machine.
No sponsor's line to hold. The lab's views are its own.
Where risk stands
From about 9 in 2019 to 44.5 today: past the line of what institutions planned for. 60 marks the edge of tolerance, 80 the edge of capacity.
Next
Scenario Atlas · Human Agency Monitor · Thought Experiment Studio · Organisation Lab · Daily Transition Brief.