Weekly Dispatch on AI, Geopolitics and Markets #14

A viral report from Citrini Research shook markets on Monday. The note outlined a scenario where rapid advances in artificial intelligence trigger what it calls a global intelligence crisis and large-scale white-collar job losses.

The argument is that the real risk is no longer overspending by big tech or narrow software disruption. It is the erosion of the economic premium attached to human intelligence itself. Framed as a June 2028 scenario, not a forecast, the report says AI-driven cost cuts across industries could lead to unemployment and financial contagion.

Markets reacted fast. The Dow fell 1.7 percent, down 822 points. The S&P 500 dropped 1 percent, and the Nasdaq fell 1.1 percent as AI fears mixed with trade uncertainty. By Tuesday, Wall Street bounced back. The S&P rose about 0.8 percent to around 6,890. The Dow gained roughly 370 points, and the Nasdaq climbed about 1 percent.

The intelligence crisis framing sounds dramatic. It is not new. Many economists and academics have warned for years that income inequality is rising as technology-driven growth becomes less tied to broad employment. I have made the same point in almost every previous weekly newsletter. AI did not start this trend, but it is accelerating it.

Growth without participation is unstable. The cake may grow, but fewer people get meaningful slices. The same corporations and individuals capture a larger share of the upside. When productivity gains flow mostly to capital owners and top-tier talent, the middle weakens. Demand becomes fragile. Politics becomes toxic. The system looks strong on paper and hollow underneath. Populism grows, authoritarianism rises.

For me, the bigger risk is not an intelligence crisis. It is a government ignorance crisis. Policymakers see the acceleration and still move slowly. There is no serious plan for redistribution, reskilling, or redesigning the flow of income in an AI-heavy economy. Weak wages and weak demand will not support endless asset inflation.

Will the future look like the Matrix or Mad Max? In a Matrix economy, AI runs production, logistics, finance, and even governance with extreme efficiency. Wealth concentrates around those who own the systems. Most people are supported but not essential. Life is stable and abundant, but human agency shrinks. In a Mad Max economy, systems break down. Supply chains fail. Energy grids weaken. Institutions lose control. Power belongs to whoever controls resources and force. Growth disappears, and survival becomes the main economic activity.

A new Federal Reserve staff paper finds that Kalshi’s regulated prediction markets are emerging as one of the most accurate sources for US macro expectations. The study compares Kalshi forecasts with Bloomberg consensus, financial derivatives, and the New York Fed Survey of Market Expectations.

The authors conclude that Kalshi’s real money contracts generate continuous probability distributions for inflation, labor data, GDP, recession risk, and Federal Reserve decisions. In several areas, including headline CPI, Kalshi outperforms Bloomberg’s economist consensus. For federal funds rate forecasts about 150 days ahead, Kalshi matches top professional forecasters and outperforms fed funds futures right before FOMC decisions.

The paper also shows that Kalshi reacts to news within minutes. It captures shifts after policy speeches and data releases that daily surveys miss. It finds that positive inflation surprises push rate expectations up more strongly than negative surprises pull them down. FOMC statements raise both the expected path of rates and uncertainty, while press conferences reduce extreme tail risks.

I am not naturally sympathetic to prediction markets. They can easily become another form of gambling, especially for young and marginalized people looking for quick wins. I also do not believe that retail wagers should guide our economic thinking. My instinct is to trust experts who have wisdom and judgement, and are accountable for their forecasts.

But this Fed paper is serious and careful. Kalshi’s performance on inflation and rate calls is hard to ignore. The speed and depth of its probability ranges show that many participants are not guessing. They are processing information fast and with discipline. This does not replace institutions. It does not make economists irrelevant. But it shows that real-time market structures can surface informed views in ways traditional surveys cannot. In uncertain times, that matters.

Nuclear talks between the United States and Iran will resume in Oman on Thursday, but expectations are low. Iran insists on continuing domestic uranium enrichment and rejects efforts to expand the talks to include missiles, regional proxies, or domestic repression.

At the same time, the United States has increased its military presence in the region. Additional naval assets have been deployed to the Persian Gulf and eastern Mediterranean, including destroyers and support vessels aimed at deterring escalation. Air activity has increased, with surveillance and strike capabilities moved closer to potential flashpoints.

Diplomats say both sides prefer negotiation over confrontation. Yet with talks stalled and military posture rising, the space for de-escalation appears to be narrowing.

I am looking at the posture and the incentives, and my view is blunt. The United States appears to be preparing for strikes on Iran, and the timeline feels short.

The scale of the military deployment is not symbolic. Carrier groups, warships, air assets. This is a real strike capacity. You do not assemble this kind of hardware without keeping the option very real.

Diplomacy looks stalled. The gaps are wide, and neither side is backing down. When talks drag on while hardware moves into place, history suggests the balance is shifting toward force.

There is also domestic politics. Presidents under pressure often turn to external action. That pattern is familiar in American history.

None of this means a strike is guaranteed. But the signals are aligned in a worrying way. A conflict with Iran would reshape the region and the global economy overnight. Energy markets would react immediately. Regional alliances would be tested. Escalation risks would be high.

My base case is that the probability of US strikes on Iran is rising fast. If talks collapse publicly, the infrastructure for action is already in place.

Kalshi and Polymarket have seen explosive growth as users bet on elections, inflation, Fed decisions, and cultural events. What began as a niche fintech experiment is now a mainstream speculative arena attracting retail traders, institutions, and regulatory scrutiny.

The core issue is simple. At what point does a prediction market become gambling? Kalshi operates under CFTC oversight, which gives it formal legitimacy. Polymarket runs on crypto rails and sits in a looser regulatory space, making it a bigger target for enforcement.

Lawmakers from both parties are questioning whether existing regulators have the authority or the will to police these platforms. Consumer advocates argue they tap into gambling instincts without offering the protections required of casinos or sportsbooks. Election-related contracts have heightened the debate's political sensitivity.

Regulatory action now looks likely. If state agencies crack down hard, innovation in event based markets could slow or move offshore. If they go soft, prediction markets could expand rapidly without clear guardrails. The trend is clear. As volumes surge and visibility grows, prediction markets are shifting from a regulatory gray zone to a regulatory battleground.

🧠 Pro Worker AI?
Daron Acemoglu draws a sharp distinction between technologies that replace workers and those that make them more valuable. He defines pro-worker AI as systems that expand human capability rather than automate it away.
https://www.nber.org/papers/w34854

👽 Are We Visited by Extraterrestrials?
Avi Loeb asks a question most scientists avoid publicly and treats it with intellectual seriousness. Whether you agree or not, the essay is a reminder that frontier inquiry requires tolerance for low probability but high impact possibilities. https://avi-loeb.medium.com/are-we-visited-by-extraterrestrials-04ea91541d7a

📊 AI Is Not Improving Productivity?
Acemoglu argues that AI’s economic impact depends less on technical breakthroughs and more on incentives and institutional choices. Productivity gains are not automatic. They reflect how systems are designed and who they are designed for. A sobering counterweight to techno optimism.
https://sloanreview.mit.edu/audio/ai-is-not-improving-productivity-nobel-laureate-daron-acemoglu/

🤖 Daniel Guetta on the Guts of AI and Why LLMs Hallucinate
A fun to watch conversation on how large language models actually work, where they fail, and what agentic AI might change. Useful for separating marketing narratives from architectural reality.
https://www.youtube.com/watch?v=Rv_rPAQ3M9Q

🌍 Iran, the US, and the Risk of War —Marc Champion
A clear argument that confrontation with Iran may be closer than markets assume. It forces readers to think through escalation risks, regional incentives, and political timing.
https://www.bloomberg.com/opinion/articles/2026-02-21/iran-us-war-is-coming-and-trump-needs-to-explain-why