The Myth of the Crowd: How Prediction Markets Reflect Power, Not Wisdom
Trump's AI Action Plan, confident conspiracy theorists, and ChimpGPT.
📖 Weekly Reads
Conspiracy theorists are overconfident — big time! — often believing they’re views are in the majority, according to new research.
Chatbots aren’t lying to us, we’re lying to ourselves. As
explains, when a chatbot produces a dishonest output, it is the result of being prompted to role-play during pre-training, and the unintended side effects of post-training human feedback.Writing is thinking. So what happens when we outsource writing to a chatbot?
Why AI personalization echoes the problems of social media and portends “a world of disempowerment that looks quite different than many might currently contemplate.” Learn the lessons and find out what we can do about it in a great essay by
.Are we misunderstanding AI like we misunderstood chimps in the 1970s? (e.g. projecting human traits without a strong theoretical framework for understanding why chimps do what they do) This fascinating paper suggests that we are, and offers steps AI research can take to be more scientifically rigorous.
The Trump Administration launched its "AI Action Plan" for "Preventing Woke AI In The Federal Government" this week. The document is unserious and deeply problematic. For example, it requires the National Institute for Standards & Technology (NIST) to "eliminate references to misinformation, Diversity, Equity, and Inclusion, and climate change." Removing these references won't eliminate data centers' environmental impacts or make technology neutral. All AI systems are biased—they're trained on human-created data!—and ignoring this fact merely disguises right-wing values as neutrality. In response, a coalition of 100+ organizations, including Data & Society (where I work), has offered an alternative called the People's AI Action Plan.
“I think, to help others have this sense of wonder, of seeing things anew, things that can catch us off guard, that break in on our small, bordered worlds.” - Ann Lamont, on the goal of writing.
🔮 Speculating on uncertainty
Lately, I’ve been writing a lot about uncertainty, including:
The trend toward increasingly uncertain and speculative lives is perhaps best captured by the rise of decentralized prediction markets.
In recent months, thousands of people have placed bets on when the LA fires would be contained, who would become the next Pope, and whether President Trump would say "tampon" in a speech. While these bets often commodify tragedy or reflect fringe interests, the phenomenon itself has become mainstream. By November last year, Polymarket — the largest prediction market — had generated over $3.68B in trading volume from 88,000 participants. Users buy tokens priced between $0 and $1, with the value representing the probability of a specific outcome. A contract trading at $.60 suggests a 60% likelihood of that outcome occurring. Proponents claim these markets simply aggregate diverse independent judgments democratically; that they represent the “wisdom of the crowds.” They're wrong.
A great new research paper analyzing nearly 500 contracts on Polymarket reveals that "prediction markets operate through information hierarchies where knowledge cascades from elite traders to broader participant circles through structured patterns."
Keep reading with a 7-day free trial
Subscribe to Untangled with Charley Johnson to keep reading this post and get 7 days of free access to the full post archives.