The most important thing about “We Must Act Now” is not what it predicts. It barely predicts anything.

The 88-word statement says AI may become radically more powerful, could transform the economy and could displace workers. Its one unqualified claim is that economists, policymakers and technology leaders must act now.

That distinction matters. The confidence is not in the forecast. It is in the need to prepare.

The 236-person core group is impressive: 16 economics laureates, leading labour economists, AI researchers, frontier-lab insiders, technology investors and people who normally disagree about almost everything. Daron Acemoglu and Joseph Stiglitz appear alongside Tyler Cowen and Alex Tabarrok. Yoshua Bengio and Yann LeCun both signed.

That breadth is the statement’s real content.

It tells us that transformative AI has moved from the edges of economics into its mainstream. The shared position is not that mass unemployment is coming. It is that the possibility is now too credible, the consequences too large and institutions too slow for uncertainty to remain an excuse for inaction.

But the agreement is also remarkably thin.

The statement names no policy, authority, deadline, funding commitment or trigger for intervention. It says nothing about ownership, taxation, competition, social protection or how productivity gains should be divided. People from major AI companies signed as individuals, but their employers made no commitments.

Its grammar is revealing too. Economists, policymakers and technology leaders are told to steer. Workers and citizens appear only as the people who might benefit from that steering. The statement calls for institutions without addressing who should have power inside them.

There is also a counting trap. The website now displays more than 900 signature records, but only 236 form the clean launch coalition. The rest are public submissions, including duplicates, blank affiliations and obvious junk. A large number on a website is not the same thing as verified expert consensus.

Current labour evidence remains mixed. There are concerning signals around exposed entry-level work, but no clear economy-wide AI employment shock yet. The statement is therefore not proof of a crisis. It is an argument for preparation before proof becomes unavoidable.

Its achievement is to make uncertainty actionable.

Its weakness is that the people already closest to AI have largely nominated themselves to do the steering.