The Daily View 7/11/2026: Superforecasters , Iran, and AI & Cheating in College

A Brown University professor suspects a majority of his class used AI to cheat on a take-home exam, highlighting the challenge of adapting assessments to the current AI landscape. The incident underscores the tension between traditional academic integrity and the widespread availability of generative AI tools.

Item 1. From Scott, “ The AI Superforecasters Are Here https://www.astralcodexten.com/p/the-ai-superforecasters-are-here ” He writes: You, personally, should not play the prediction markets in the AI future. You’ll be competing against smarter-than-top-human superforecasters that can spend subjective weeks cogitating on every single question, and you will definitely lose. To the contrary, it would actually be easier, at least initially. You can just copy the forecasters’ trades. This would work until it’s arbitraged-away. I think the superforecasting concept sorta outlived its usefulness. There was a lot of hype about this from 2015-2020 after being popularized by Philip E. Tetlock, garnering books and mentions on podcasts, but a decade later, not much has come out of it. I have yet to see much evidence anyone can make correct predictions consistently for practical or actionable matters, like stock market performance or which sectors will lead and so on. It’s like,” tell me which sectors will lead next year. and the following year. ” Repeated enough times with enough participants and you will get outcomes indistinguishable from randomness. The strongest evidence superforecasting doesn’t work, is the fact hedge funds don’t hire superforecasters. Even if hedge found tried to downplay this for competitive reasons, word would still eventually get out that funds are hiring forecasters, but there is no evidence to suggest so. Hedge fund want actionable advice and knowledge that can be applied within the window of a cycle, typically within a year or two. Just making a bunch of forecasts that may take a decade to play out it less useful. Item 2. Speaking of forecasting, I keep being right about Iran. A month ago, people were tweeting that Trump had “won,” and I was thinking, “Won what? The war he started that Israel won’t let him finish?” Here is what I wrote on June 19th, from the post https://greyenlightenment.com/2026/06/19/the-iran-deal-that-wont-happen/ , “The Iran Deal That Won’t Happen:” Even if something is eventually finalized, in a few months, maybe only weeks, it’s a virtual certainty both sides will soon be accusing each other of violating parts of the agreement. It won’t be long before Trump is again threatening additional strikes. Sure enough, Iran strikes are back “ on https://www.bbc.com/news/articles/cz75zjj5wp8o ” and the ceasefire deal is “ off https://thehill.com/policy/defense/5960413-us-iran-strikes-intensify-donald-trump-ceasefire-over/ “. My original forecast https://greyenlightenment.com/2026/03/25/the-strait-of-hormuz-will-remain-closed-possibly-until-2028/ from March 2026 of the strait remaining in some state of being blockaded until at the earliest 2028 is looking more and more correct. As long as Trump is in office, Iran will never give up that bargaining chip. The war is clearly not popular, and high inflation is arguably the biggest threat to the economy now. Iran will try to keep oil prices high to hurt Trump’s approval ratings so that his successor loses, after which Iran will reopen the strait fully. Item 3. I’m sure everyone has seen this, “ Brown Professor Suspects Majority of His Class Used AI to Cheat https://www.insidehighered.com/news/faculty/learning-assessment/2026/07/08/brown-professor-suspects-most-his-class-used-ai-cheat ” Friends don't let friends continue to give exams and papers that are not adapted to the current AI landscape this is AFTER 27 bailed so the reality is even worse PS Props to students 1, 22, and 31 pic.twitter.com/3LPj66YuQT — Brendan Nyhan @BrendanNyhan on 🟦☁️ @BrendanNyhan July 8, 2026 The idea of a take-home exam is an oxymoron. Why is anyone surprised cheating is going on. Even without AI, Google exists. Professors don’t want to proctor exams, and assume that students will uphold some strict moral code, when in reality, they are paying for a credential rather than an education. If success at school is measured by producing right answers, and as we know AI is exceptionally good at doing exactly that, then being rational actors, students will use AI. If professors, who already have extremely flexible and generous work accommodations, at a bare minumum cannot proctor exams, they should be replaced by staff who will. It’s not as if there is a labor shortage in academia. The problem with education is that it’s inherently circular, by reproducing accepted knowledge. Something is “correct or true” because enough important people have deemed it so. So is there any fundamental difference between an AI or a human producing the same answer if it’s correct? This is why common arguments that the widespread use of AI will lead to unqualified “cheaters” being hired e.g. law school, engineering, or medicine are unfounded, because it make no difference who or how an answer is produced either AI or humans , provided there is some consensus mechanism of what the correct answer is? If the knowledge of making a bridge or surgery exists, what difference does it make if humans or AI applies it?