AI Solved 35% of Theory Problems by Mid-2026

Today’s essential intelligence on markets, energy, AI and geopolitics.

Key takeaways:
• Political transformations and geopolitical challenges
• Economic stress tests and global resilience
• Environmental crises and climate impacts
• AI’s expanding influence in science

MIT study says AI may soon be the default tool for theoretical science
MIT researcher Neil Thompson’s latest study, reported by Wired, finds that several fields are about to be upended by AI. US proposes releasing part of frozen Russian assets into a Washington-run vehicle
The US has proposed that European states free a portion of the Russian state assets blocked since the invasion of Ukraine.

Read the full newsletter: https://thegist.online/2026-10-15-mits-neil-thompson-found-ai-models-solving-en/
Subscribe free: https://thegist.online/subscribe-to-the-gist/?utm_source=podcast-en&utm_medium=show_notes

Listen to this episode

Transcript

JOHN: Welcome to The Gist. It is Thursday, October 15th, 2026. I am John.

MARY: And I am Mary. We are your smart friends on the go. Let us get right into the news.

JOHN: We start with The Gist View. Today we are looking at the business of science. Or rather, the automation of it.

MARY: A new MIT study just dropped in Wired. It says AI models are now solving about 35 percent of new theoretical science problems. Up from just 2 percent two years ago.

JOHN: That sounds like a massive breakthrough. But there is a catch. The MIT team had to test this offline using OpenAI’s proprietary models, from o1 up to GPT-5.5.

MARY: And who checks if these AI answers are actually right? Human experts.

JOHN: Exactly. Let us look at the incentives here. Two weeks ago, OpenAI dumped over 700 manuscripts online from an internal model. They kept the model secret. They only published the results they liked.

MARY: This creates a huge power imbalance. The AI labs get the headlines for mass-producing science. But the academic referees bear the cost. They do the unpaid labor of checking the math. And they cannot even test the model that generated it.

JOHN: A Princeton advisory group recently called out this behavior. They criticized AI companies for relying on secret internal models. OpenAI basically said they do not have to listen.

MARY: Compute is power. Theoretical science is cheap to automate because you do not need a physical lab. You just need massive servers. Money and prestige will flow to whoever owns the biggest models. Universities without that access will just become dependent clients.

JOHN: Moving to the Global Overview. Let us look at energy and resilience.

MARY: The International Monetary Fund, or IMF, has a new update on Asia. They say Asian economies are handling Middle East fuel shocks much better than expected.

JOHN: It is all about supply chains. The IMF notes this resilience is uneven. Countries with diversified suppliers and large reserves are fine. But if you rely on a single pipeline or single shipping route, you are exposed.

MARY: Speaking of exposure. Air quality in Singapore took a nosedive this week. The cause is wildfires from plantations in neighboring Indonesia.

JOHN: Singapore is now weighing legal action against the companies responsible. Bloomberg reports they will use their courts to go after corporate assets.

MARY: Why courts? Because the Indonesian government struggles to police the remote land where the fires actually start. But Singapore is a financial hub. If an errant company holds money or assets there, Singapore has the leverage to squeeze them.

JOHN: Over in Eastern Europe, a new kind of target is emerging. Ukraine is striking Russian data centers.

MARY: Politico reports these strikes aim to cut off military planning. But the servers hold civilian data, too.

JOHN: This creates a major gray area in international law. A data center is a dual-use asset. The people hosting cloud data in contested regions now face very real physical risks.

MARY: Down in France, the presidential race is making bond investors nervous. Marine Le Pen is leading the polls. She wants to slash billions of euros from France’s payments to the European Union.

JOHN: She wants to reduce the powerful European Commission to a mere secretariat.

MARY: Meanwhile, on the left, Jean-Luc Mélenchon is also a real contender. He wants to cancel France’s national debt.

JOHN: Both challengers are taking aim at the exact financial structures that bind Paris to Brussels. If either wins, the flow of resources in Europe shifts dramatically.

MARY: That brings us to The European Perspective. Let us talk about frozen Russian money.

JOHN: When Russia invaded Ukraine, Europe froze billions in Russian state assets. Now, Washington wants a piece of the action.

MARY: US envoys Steve Witkoff and Jared Kushner just proposed a new plan. They want Europe to move a chunk of that money into a vehicle run by Washington. The stated goal is to use it for peace talks.

JOHN: European delegates flat-out rejected it. This is a pure power play. The money is sitting in European institutions. That means Europe holds the legal custody.

MARY: Custody is leverage. Moving those funds under US oversight transfers the power over reconstruction and sanctions straight from Brussels to Washington. Europe is not giving up that leverage.

JOHN: Here in Germany, a quick update on health insurance. Health Minister Carsten Linnemann says the extra contribution rate will stay frozen at 2.9 percent for 2027.

MARY: He is calling it a win. But a flat rate on paper does not fix the actual costs. The system’s finances are highly strained. Freezing the rate just drains the reserve funds and pushes the hard budget choices to next year.

JOHN: And daily costs keep rising. The ifo Institute tracks what German companies plan to do with their prices. Their index just jumped to 22.7 points.

MARY: Energy is the culprit. Gas and power prices hit their highest point since late 2022. Economists expect German inflation to top 3 percent soon, right as winter utility bills hit the mail.

JOHN: Let us pan out to the global economy. Why are some countries rich and others poor? For a long time, economists blamed the gap on how many years people spent in school.

MARY: But a new study from the CEPR—a leading economic research network—changes the math. They found that up to 70 percent of the income gap between countries comes down to measured skills.

JOHN: Not just years in a classroom. It is about the actual quality of the education and the value of real work experience.

MARY: This totally shifts the incentives for global aid. Building a school is easy. Training highly effective teachers is hard. But this data shows training quality is what actually moves the needle on national wealth.

JOHN: Finally, a quick political note from Berlin. Exploratory talks for a left-wing coalition are hanging by a thread today.

MARY: The SPD, the Greens, and the Left Party are arguing over antisemitic factions within the Left. It is a messy dispute, and it comes at a terrible time for the ruling SPD, who are facing awful poll numbers ahead of their December conference.

JOHN: That is the daily temperature. Today is a story of bottlenecks and leverage. Whether it is AI labs hoarding compute to command scientific prestige, Singapore using its financial courts to fight foreign smog, or Brussels refusing to hand Washington the keys to frozen Russian cash. In every case, whoever controls the choke point dictates the terms of the future.

MARY: And that is The Gist for today. If this breakdown helped you start your Thursday, we would love to keep the conversation going. You can get our daily insights delivered to your inbox for free—just tap the subscribe link right in the show notes. Talk to you tomorrow!


The Gist is an independent daily digest: AI-curated, human-directed, unapologetically liberal (how it’s made). Hundreds of sources, only what matters. Subscribe free or listen to the podcast.