In Brief
The AI Futures Project, the group behind the widely debated AI 2027 forecast, published AI 2040: Plan A, a 90-page scenario arguing the world should strike an international deal to delay superintelligence until 2040.
It is explicitly a recommendation, not a prediction, and Paul Roetzer says its near-term setup is the part Americans cannot afford to ignore.
What Happened
The AI Futures Project, a research group led by former OpenAI researcher Daniel Kokotajlo, published AI 2040: Plan A, a 90-page scenario laying out the team's positive vision for how humanity reaches superintelligence. Superintelligent AI outperforms the best human experts at virtually every task, without catastrophe.
This is the same group whose AI 2027 scenario forecast that AI could automate AI research itself within a couple of years, ending in either AI takeover or an extreme concentration of power. The follow-up flips from warning to advice: "We recommend an international deal to avoid a dangerous race to superintelligence," the authors write. The deal involves total research transparency for AI research and development, so nations can enforce guardrails, with multiple companies across multiple countries scaling slowly and safely together instead of racing in secrecy.
"Plan A is our current best guess, but hopefully a better plan will exist before it's too late," Kokotajlo wrote on X. "We hope that the best ideas from Plan A will be adopted and the worst discarded."
SmarterX founder and CEO Paul Roetzer examined why a speculative 90-page scenario earned a main-topic slot on Episode 225 of The Artificial Intelligence Show.
The Key Numbers
90 - Pages in the AI 2040: Plan A scenario
2040 - Plan A's target arrival year for superintelligence
165 million - Human workers in the scenario's 2027 US workforce, alongside millions of AI agents
2x - The 2028 scenario's data center costs vs. the entire US military budget
5 - Plans the scenario puts in front of the 2028 presidential candidates
A Political Firestorm Is Coming
The 2027 scenario isn't far-fetched. The report opens with a hypothetical 2027 in which "America has two workforces now. The first is people, 165 million of them. The second is AI agents, millions of copies spun up and shut down every hour working around the clock at superhuman speeds." AI agents are AI systems that complete multi-step tasks on their own. In the scenario there is still no recursive self-improvement, the loop where AI does its own AI research and makes each generation smarter, but the labs are getting closer and the strongest coding AIs refuse to help competitors. Congress wakes up and passes a hypothetical AI Transparency Act of 2027 that changes little. "None of this is far-fetched at all," says Roetzer. "This is like next year."
The 2028 election scenario follows the same logic. In this scenario, AI is the biggest topic of the presidential race, a prediction Roetzer fully agrees with. Data centers under construction cost twice the entire US military budget. Most white-collar professions now heavily involve managing AI agents, a shift Roetzer says is absolutely going to happen. And power concentrates in the president plus a handful of tech CEOs, which he notes is already underway.
Plan A is one of five options, and Roetzer doubts it is the viable one. The candidates converge on competing plans: Plan A, a verified slowdown through international cooperation; Plan B, fight China with a US-led coalition; Plan C, burn the lead with strong safety regulation; Plan D, race to superintelligence with light-touch rules; and Plan S, shut it all down with a global moratorium. On Plan A itself, Roetzer says: "Sounds great, does not seem like a viable option in my opinion." He adds that anything beyond 2028 is completely guesswork. The near-term choice among these plans is the part he takes seriously.
"It's going to play a role in the midterms in '26. It will dominate the 2028 presidential election. There's no way it can't, because the implications are so vast across the economy, energy, where we're getting energy from, where we're building data centers, what foreign entities we're allowing to invest in U.S. companies."
— Paul Roetzer, founder and CEO of SmarterX, Episode 225 of The Artificial Intelligence Show
SmarterX Take
You do not have to buy the 2040 forecast to get value from this document. Treat the 2027 and 2028 chapters as a planning horizon. The two-workforce framing is a practical lens: if your organization's capacity is soon measured in people plus AI agents, hiring plans and budgets built only around headcount are already out of date.
AI policy is also about to become a business variable. Decisions on energy, data centers, chip exports, and foreign investment will shape costs and competition for every company, and reading documents like this one builds awareness to see those fights coming.
What to Watch
The 2026 midterms are the first test of AI as a campaign issue. Roetzer's conviction is that the political fight arrives on schedule whether or not the rest of the scenario does. "It is going to be very, very important that people understand these issues going into 2028," he says.
Engagement from labs and governments will show whether Plan A gets traction. If policymakers start referencing verified slowdowns, research transparency, or international coordination, the scenario will have done its job as an agenda-setter, even if 2040 never happens as written.
What 2,100+ Professionals Expect AI to Do to Jobs
The workforce already senses the future this scenario describes. Seventy-one percent of professionals believe AI will eliminate more jobs than it creates over the next three years, versus just 13% who expect net job creation, according to the 2026 State of AI for Business Report. Yet only 20% express active concern about their own role.
The full report, based on 2,100+ responses from professionals across roles, functions, and industries, maps how the workforce actually feels about AI's impact on jobs, careers, and organizational readiness. Read the full report →
Mike Kaput
Mike Kaput is the Chief Content Officer at SmarterX and a leading voice on the application of AI in business. He is the co-author of Marketing Artificial Intelligence and co-host of The Artificial Intelligence Show podcast.