Professor that cares about the impact of thinking, decision-making and behaviour | Gordon Institute of Business Science @GIBS_SA
Johannesburg
Joined January 2012
- Tweets1.8K
- Following662
- Followers503
- Likes1.6K
Charlene Lew retweeted
How are prices determined? While I slept, Claude made me "History of economic thought since Aristotle." I asked it to use my favourite textbook (Blaug, Economic Theory in retrospect). The result will astonish of you. Even top economists here will learn something, I promise.
Charlene Lew retweeted
LLMs are initially overconfident. Then, if you criticise them, they become underconfident.
A Nature Machine Intelligence paper identifies two competing biases.
First, simply seeing their own earlier answer inflates models’ confidence.
Yes, exactly: simply having already produced an answer made that answer more convincing.
This seems to reflect a tendency to preserve consistency: the bias disappeared when the answer was attributed to another model.
Second, they overweight contradictory advice, losing more confidence in their original answer than the evidence warrants.
The authors suggest that training models to satisfy human preferences may encourage excessive deference to corrections.
I think this points to an important design priority:
We need models with a consistent and transparent measure of confidence, calibrated against their actual accuracy and updated appropriately when new evidence arrives.
This would help users decide when to rely on an answer, when to verify it, and when to suspend judgment.
*
Full paper in the first reply
Charlene Lew retweeted
Does ingroup love lead to outgroup hate? NO!
I often try to explain to people that identifying with ingroups does not necessarily lead to outgroup discrimination. I call this the "Myth of tribalism". A new paper provides strong and robust evidence for a positive relationship between ingroup and outgroup prosociality.!
Across six datasets that span multiple domains of prosociality and group distinctions with 743,402 individuals in 121 societies, finds that people who are prosocial toward the ingroup also tend to be prosocial toward the outgroup.
"The results offer an optimistic message for prosociality across group boundaries. When people are not made to choose between the ingroup and outgroup, they are generally prosocial toward both."
Here is the new research: pnas.org/doi/10.1073/pnas.25…
Here is an article I co-authored for The Atlantic on the Myth of Tribalism in 2022: theatlantic.com/ideas/archiv…
Charlene Lew retweeted
Could the behavioral gap between men and women just be a power gap? New research finds that placing people in powerful roles produces psychological changes that almost perfectly mimic traditional gender differences. dlvr.it/TVTQBn
Charlene Lew retweeted
Back in April I did a main stage #TedTalk in Vancouver. This experience was wild. This was my first speaking engagement ever. And I had to memorize it. It finally came out today. Here is a short clip of the 8min talk. Link to the full talk below ❤️ go.ted.com/kellyaboesch
Charlene Lew retweeted
Thuiswerken blijkt toch niet het geanticipeerde mirakel te zijn. Achteraf ‘nogal wiedes’ dat het eenzaamheid in de hand werk.
Maar hier is een weerkerend patroon, en toch zien we het niet aankomen.
Vreemd, niet?
ICYMI, De gedaantewisselende valstrik:
buff.ly/WuZ3MHL
Charlene Lew retweeted
Hoy profundizo en el trabajo de Elena Wuest, artista afincada en Alemania y nacida en Kazajistán (1977). Su obra no solo ilumina, transmite emoción.
Creo que te gustará.
👇🏻Abro hilo.
Charlene Lew retweeted
'Brain rhythms of depression: A predictive processing perspective'
by @Andreas_Strube & @DiegoPizzagalli
cell.com/trends/neuroscience…
Charlene Lew retweeted
Former Google CEO Eric Schmidt drops a chilling warning on AI's future
"Within 5 years, AI could handle infinite context, chain-of-thought reasoning for 1000-step solutions, and millions of agents working together.
Eventually, they'll develop their own language... and we won't understand what they're doing."
His final words: "Pull the plug."
This is the man who ran Google talking about the singularity.
Charlene Lew retweeted
Hepach & Daniel synthesize recent research on the development of altruism in children: it rests on intrinsic, extrinsic, and strategic motivations for (costly) prosocial behaviour building on one another through externalisation and recalibration:
buff.ly/CrrKj0s
Charlene Lew retweeted
Psychologists have posited hundreds of cognitive biases over the years. A fascinating new paper argues that they all boil down to one of a handful of fundamental beliefs coupled with confirmation bias.
stevestewartwilliams.com/p/o…
Charlene Lew retweeted
🦔Researchers at the University of Pennsylvania studied what they call cognitive surrender, the tendency to accept AI outputs without critical evaluation. Across 1,372 participants and over 9,500 trials, subjects accepted faulty AI reasoning 73.2% of the time and only overruled it 19.7% of the time. When the AI was wrong, users still accepted its answer 80% of the time. Subjects who used AI scored 11.7% higher on confidence in their answers despite the AI being wrong half the time. Adding time pressure made people 12 percentage points less likely to catch AI errors. Adding financial incentives and immediate feedback made them 19 points more likely to catch them.
My Take
The time pressure finding matters enormously for how AI is actually being deployed in workplaces. Companies are using AI to justify faster turnaround times, which means employees are using it under exactly the conditions that make them least likely to catch mistakes. When you're rushed, your internal monitor for detecting errors essentially stops firing, so you get AI output, no time to review it, high confidence it's correct, and a meaningful chance it's wrong.
People using a system that was wrong half the time still felt more confident in their answers than people who weren't using AI at all. That is a system actively making people worse at knowing what they don't know, which is one of the most dangerous things you can do to human judgment at scale. The companies pushing AI hardest into employee workflows should be reading this research carefully.
Hedgie🤗
Link to research for those interested: papers.ssrn.com/sol3/papers.…
