Stef Götz
6.3K posts
Stef Götz
@Tangl3dBank
social neuroendo | evopsy, culture,&their X| Member of the reading public | Aspirant to the chattering class | culturally encoded | Magic Dirt Enthusiast
Wharton’s latest AI study points to a hard truth: “AI writes, humans review” model is breaking down Why "just review the AI output" doesn't work anymore, our brains literally give up. We have started doing "Cognitive Surrender" to AI - Wharton’s latest AI study points to a hard truth: reviewing AI output is not a reliable safeguard when cognition itself starts to defer to the machine.when you stop verifying what the AI tells you, and you don't even realize you stopped. It's different from offloading, like using a calculator. With offloading you know the tool did the work. With surrender, your brain recodes the AI's answer as YOUR judgment. You genuinely believe you thought it through yourself. Says AI is becoming a 3rd thinking system, and people often trust it too easily. You know Kahneman's System 1 (fast intuition) and System 2 (slow analysis)? They're saying AI is now System 3, an external cognitive system that operates outside your brain. And when you use it enough, something happens that they call Cognitive Surrender. Cognitive surrender is trickier: AI gives an answer, you stop really questioning it, and your brain starts treating that output as your own conclusion. It does not feel outsourced. It feels self-generated. The data makes it hard to brush off. Across 3 preregistered studies with 1,372 participants and 9,593 trials, people turned to AI on over 50% of questions. In Study 1, when AI was correct, people followed it 92.7% of the time. When it was wrong, they still followed it 79.8% of the time. Without AI, baseline accuracy was 45.8%. With correct AI, it jumped to 71.0%. With incorrect AI, it dropped to 31.5%, worse than having no AI. Access to AI also boosted confidence by 11.7 percentage points, even when the answers were wrong. Human review is supposed to be the safety net. But this research suggests the safety net has a hole in it: people do not just miss bad AI output; they become more confident in it. Time pressure did not eliminate the effect. Incentives and feedback reduced it but did not remove it. And the people most resistant tended to score higher on fluid intelligence and need for cognition. That makes this feel less like a laziness problem and more like a cognitive architecture problem.
A new paper on correlates of zero-sum thinking just came out and it has everything: race, sex, politics, class, mobility, and even deep roots! Here are my favorite graphs from the paper. But first: how was zero-sum thinking measured? The answer is as a set of four questions on whether things are zero-sum with respect to ethnicity ("If one ethnic group becomes richer, this generally comes at the expense of other groups in the country"), citizenship ("If those without American citizenship do better economically, this will generally come at the expense of American citizens"), trade ("In international trade, if one country makes more money, then it is generally the case that the other country makes less money"), and income ("If one group becomes wealthier, it is usually the case that this comes at the expense of other groups"). Endorsement of these ideas is considered zero-sum thinking. Because these can feel political, you might think that would compromise the results. And true! Measurement invariance wasn't tested, but removing mechanically-related questions didn't seem to change this paper's findings much. Onto the graphs! The first one I liked was on the demographics of zero-sum thinking. It's a young and middle-aged person's game, but it's also a game for Hispanics and Blacks, but not Asians, for Democrats and not Republicans, for urbanites, somewhat for ruralites, and not as much for suburbanites, and there are U-shaped relationships with income and education. There are lots of findings in the break-downs of these categories, like that Democrats who voted for Trump were often highly zero-sum thinkers, or that zero-sum thinking is simultaneously related to - The belief that luck trounces effort - The perception that mobility is high - Universalist values - A belief in the importance of tradition - Generalized trust A second finding I found extremely interesting was that people who experienced more growth in the first twenty years of their lives had less zero-sum values. Because of the correlation between growth and zero-sum thinking over time and compositional changes that covary with those changes, it's important to do some post-stratification to see if this result really holds up. If it does, it has fascinating implications. The paper is really chock-full of fun facts, like that, globally, right-wingedness is related to less zero-sum thinking, but in some countries, the relationship is nullified or reversed. Another finding was that being anti-immigrant and pro-redistribution was related to zero-sum thinking among Democrats, and even more strongly, among Republicans. Yet another finding was that parental, grandparental, and great-grandparental mobility was negatively related to zero-sum thinking. A more immigrant-focused finding was that later-generation immigrants are closer to non-immigrant levels of zero-sum thinking. That is, they become more zero-sum! More likely there's selection at play, but regardless, immigrants are less zero-sum and this held up in the 2nd and 3rd generations, too. It was also found that county foreigner shares were unrelated to zero-sum thinking in respondent's generation or their parent's generation, but they were negatively related in their grandparent's generation. Another intergenerational transmission of values question had to do respondents' self-identification of having ancestors who experienced different bouts of slavery. The descendants of African slaves, Holocaust survivors, indentured servants, interned Japanese Americans, and enslaved Amerindians were more likely to be zero-sum thinkers. The same was not true for the descendants of prisoners of war. Unlike with immigration, the zero-sum correlates of enslavement seem more robust. For example, a person's county enslaved share in 1860, their parent's county enslaved share in 1860, and their grandparent's county enslaved share in 1860 all correlated with zero-sum sentiment. This remained true for people, their parents, and their grandparents if they moved out of the American South! Additionally, these findings also held true for each level when it came to county Confederate culture. In other words, the transmission of values, even with controls for demographics, state, and race was robust! This study paints a vivid picture of the correlates of zero-sum thinking in the present day, internationally, and with respect to their roots in the deep past. I definitely recommend reading it! Go check it out: nber.org/papers/w31688
What happened with Darwin was the emergence of an axiom of temporalism in which holds the trends we see now must be responsible for the patterns of yesterday, uniformitarianism. This was largely popularized by Darwin's primary confidant, Charles Lyell. x.com/i/status/19886…
