AI productivity myths, Houthi oil risks, and Ukraine's wartime democracy
Nonrival
Expert analysis from think tanks and academia, on public policy, economics, and technology.
Human experts. AI summaries.
The AI Productivity Boom May Be a Statistical Mirage — and Betting on It Could Cost Trillions
csis
- U.S. labor productivity has surged since 2022, and optimists — including the Congressional Budget Office — are attributing it to AI, but a closer look at the data suggests it's mostly a routine post-pandemic adjustment as firms shed workers they over-retained during COVID.
- The industries that use AI most intensively show no special productivity gains; instead, the sectors that hoarded labor the most are the ones now shedding it fastest, which is the signature of a cyclical correction, not a tech revolution.
- If the productivity boost proves temporary, official forecasts will be off by as much as $2.2 trillion in federal borrowing over the next decade — a serious risk at a moment when the U.S. faces enormous capital demands for defense, infrastructure, and debt service.
Ukrainians Are Protesting in Wartime Because They Refuse to Separate Winning the War from Building a Democracy
the_conversation
- New polling from Ukraine finds that even amid a grinding war, large majorities of Ukrainians want anti-corruption agencies and press freedom protected — and over a third rank government corruption as their country's biggest problem, rivaling Russian aggression.
- The largest domestic protests since the 2022 invasion erupted after the dismissal of a popular, tech-savvy defense minister who championed drone warfare over Soviet-style attrition — tapping into a generational divide over how to fight the war.
- The data suggests Ukrainians see the war not just as a military struggle but as a fight to become a European democracy, which makes accountability at home feel inseparable from victory abroad.
Trump's Declassified Election Documents Are Not the Same as Proof
the_conversation
- A former senior CIA official explains that declassifying a document only changes who can see it — not whether it's true, current, or representative of the full intelligence picture.
- Intelligence agencies deliberately preserve unreliable and later-discredited reports, meaning the government's files almost certainly contain documents that appear to support nearly any serious allegation — alongside the analysis that determined how much weight they deserved.
- When reviewing any declassified release, readers should ask whether they're seeing raw reporting or a finished assessment, and whether the release includes the context analysts used to evaluate it.
In the News
The Houthis Are Threatening Saudi Oil Exports — and This Time the Stakes Are Even Higher
atlantic_council
- The Houthis have announced a blockade of Saudi Arabia, threatening oil flows through the Bab el-Mandeb strait, which carries about 7% of global supply — a serious blow given that the Strait of Hormuz is already disrupted by an ongoing Iran war.
- The key question is whether this is a genuine military escalation or a negotiating tactic to extract concessions from Riyadh, such as formal recognition of Houthi control over northern Yemen.
- Even if it's posturing, the fragile regional environment — with Iran actively stoking the conflict — means the situation could spiral into prolonged war that further destabilizes Gulf oil markets.
Analysis
AI Analysis of 15,000 Sustainability Reports Finds They've Gotten Worse, Not Better, Over Time
harvard_corpgov
- Researchers used large language models to analyze over 15,000 corporate sustainability reports spanning 25 years, finding that as reporting became more widespread, reports actually became less specific, less quantitative, and fluffier — even among companies that had been reporting for years.
- Voluntary frameworks like GRI, SASB, and TCFD show mixed results: firms stack multiple frameworks simultaneously (over 80% of GRI adopters also use SASB), suggesting credential-collecting more than genuine improvement, though the frameworks do appear to influence what companies actually write in different ways.
- The authors argue the real problem isn't voluntary vs. mandatory disclosure, but that vague standards let companies check the box on whether they report without accountability for how well they report — and that clearer, more specific requirements are needed to move substance, not just adoption.
Also Worth a Look
- Almost No Democracy Puts Its Living Leader on Currency — Trump Wants to Be the Exception (the_conversation)
- AI Is Already Reshaping the Legal System, But Nobody Really Knows How Often It Gets Things Wrong (stanford_hai)
- AI Companies Can't Self-Regulate Open-Source Models — and That's a Problem No One Has Solved (hoover)
- A Delaware court says Porsche may have secretly strangled a startup it invested in to kill a rival's technology (harvard_corpgov)
- New York's New Power Line from Canada Proved Its Worth During a Heat Wave — Then Broke Down (eia)
- A Federal Court Just Blocked Colorado's Drug Price Cap, and the Ruling Could Hobble Every State Trying to Do the Same (the_conversation)
- The Science Is Clear: Testosterone Doesn't Make Men Tougher or More Aggressive (the_conversation)
Don't miss what's next. Subscribe to Nonrival: