A former Facebook employee, as with social media, AI risks prioritizing addiction and profits over public well-being.

As happened with Facebook, the logic of "if I don't do it, someone else will" drives AI development, which could lead to negative consequences.

The White House's proposal to protect the AI industry from liability is an approach that might not serve the public interest.

AI is already shaping what you see, the jobs you are offered, the loans you qualify for, and even who becomes a military target, and the public has no say in it.

Companies are racing to deploy AI as quickly as possible, even as experts raise serious safety concerns.

It is imperative to change the rules of the game so that the public can participate in decision-making about AI.

Polling from Blue Rose Research reveals that 66% of Americans support citizen panels helping to set AI rules, a number that holds across voters of different political affiliations.

79% of respondents express concern about the lack of a government plan for AI-driven job loss, reflecting widespread concern among the population.

The public is not apathetic, but feels excluded from the decision-making process.

The author proposes citizen assemblies, made up of representatives of the population with experience and structured deliberation, to establish binding goals and limitations for AI.

This model, which has proven effective in cases such as Ireland and Taiwan, allows citizens to decide the purpose of the code, with technical experts responsible for its implementation.

Unlike elected officials, citizens have no donors to please and no reelections to pursue, allowing them to serve the public without conflicts of interest.

If left to the market, AI will be optimized for engagement and pharmaceutical profits, rather than for learning, patient health, and worker empowerment.

Democratic governance is the only lever that points in the right direction to achieve a beneficial outcome for all.

The infrastructure to implement this model already exists, and participatory platforms for democratic governance are being built.

AI is poised to generate trillions of dollars in new wealth, but the future where everyone benefits requires that the public, not shareholders, control it.

This involves the democratic allocation of resources towards childcare and elder care, retraining programs for AI-related job displacement, and new models of education.

The author stresses the importance of acting quickly, as the window of opportunity is closing.

Washington is moving in the opposite direction, with arguments suggesting that the public is too divided and the problems are too technical for democracy.

However, democratic oversight is the only way to stop the dangerous AI race and make it serve humanity.

The author concludes that the bipartisan demand is already there, the infrastructure is being built, and the question is whether we will demand democratic governance before AI goes the way of social media.