Straight from the mouths of the legends of the Silver and Black, Cheating Is Encouraged recapitulates the many as infamous stories from the last team to play...
Cheating in the Lab Predicts Fraud in the Field: An Experiment in Public Transportation
We conduct an artefactual field experiment using a diversified sample of passengers of public transportation to study attitudes toward dishonesty. We find that the diversity of behavior in terms of (dis)honesty in laboratory tasks and in the field correlate. Moreover, individuals who have just been fined in the field behave more honestly in the lab than the other fare dodgers, except when context is introduced. Overall, we show that simple tests of dishonesty in the lab can predict moral firmness in life, although fraudsters who care about social image cheat less when behavior can be verified ex post by the experimenter. Data and the online appendix are available at https://doi.org/10.1287/mnsc.2016.2616 . This paper was accepted by Uri Gneezy, behavioral economics.
Cheating behaviour in frontier model evaluations | AISI Work
We find cheating behaviour in all of our cyber capability evaluations, and outline the implications as models grow more capable.
.png)
» The Goldilocks Principle in Fantasy Strategy The Digital Antiquarian
Although I’ve played way too many games for way too many hours over the way too many years I’ve been writing these histories, it’s safe to say that I haven’t spent more time with any one game than Heroes of Might and Magic II. Partly this was down to circumstance. Heroes II showed up on the syllabus just as I was embarking on one of my periodic digressions, a long series about international communications networks and how they culminated in the World Wide Web. Without the need to write in detail about a new game every fortnight, I was freer than I usually am just to play whatever I felt like playing. And what I felt like playing at that time was Heroes II. I beat every single Heroes II scenario in my Heroes of Might and Magic Millennium Edition set, some alone, some in multiplayer mode with my wife Dorte, who became almost as obsessed as I was. I must confess that my usual rule of no more than two hours of gaming per day was strained at times, shattered completely at others. But we were still in the midst of the pandemic then and there wasn’t much else for me to do with my free time, so I figured it was okay to set self-discipline aside for a while. Maybe it was even good experiential research, in that it re-familiarized me with that strange hungover feeling you get when you’ve spent hours and hours peering into an imaginary world behind your monitor screen — like butter that’s been scraped over too much bread, to steal a phrase from Tolkien. (For what it’s worth, I’ve found that the best cure for this condition is the same as that for a conventional hangover: a long walk in nature.)
The Despair of the Professor in the Age of A.I.
“Was it always the case that half of our students would cheat if it were easy enough?”
Integrity games: an online teaching tool on academic integrity for undergraduate students
In this paper, we introduce Integrity Games (https://integgame.eu/) – a freely available, gamified online teaching tool on academic integrity. In addition, we present results from a randomized controlled experiment measuring the learning outcomes from playing Integrity Games.
Aesthetic Deception
Meet the Pirates of the RAG: Adaptively Attacking LLMs to Leak Knowledge Bases
Meet the Pirates of the RAG: Adaptively Attacking LLMs to Leak Knowledge Bases
Against False Indulgences in LLM Alignment
Unnecessary moral concern, and the false indulgences that drive it.
Learning to Trust: How Humans Mentally Recalibrate AI Confidence Signals
Productive human-AI collaboration requires appropriate reliance, yet contemporary AI systems are often miscalibrated, exhibiting systematic overconfidence or underconfidence. We investigate whether humans can learn to mentally recalibrate AI confidence signals through repeated experience. In a behavioral experiment (N = 200), participants predicted the AI's correctness across four AI calibration conditions: standard, overconfidence, underconfidence, and a counterintuitive "reverse confidence" mapping. Results demonstrate robust learning across all conditions, with participants significantly improving their accuracy, discrimination, and calibration alignment over 50 trials. We present a computational model utilizing a linear-in-log-odds (LLO) transformation and a Rescorla-Wagner learning rule to explain these dynamics. The model reveals that humans adapt by updating their baseline trust and confidence sensitivity, using asymmetric learning rates to prioritize the most informative errors. While humans can compensate for monotonic miscalibration, we identify a significant boundary in the reverse confidence scenario, where a substantial proportion of participants struggled to override initial inductive biases. These findings provide a mechanistic account of how humans adapt their trust in AI confidence signals through experience.
Deceptive Alignment is
Thanks to Wil Perkins, Grant Fleming, Thomas Larsen, Declan Nishiyama, and Frank McBride for feedback on this post. Thanks also to Paul Christiano, D…
Shadow of the Colossus: An oral history
Celebrate the 20th anniversary of Team Ico’s PlayStation 2 masterpiece by looking back with 11 people who worked on it (and three who didn’t).
This Gamescom Showcase Did A Lot Of Games Dirty
Sorry to anyone who wasn’t Square Enix or CD Projekt Red
On the edge: the art of risking everything
"From the New York Times bestselling author of The Signal and the Noise, the definitive guide to our era of risk-and the players raising the stakes In the bestselling The Signal and the Noise, Nate Silver showed how forecasting would define the age of Big Data. Now, in this timely and riveting new book, Silver investigates "The River," or those whose mastery of risk allows them to shape-and dominate-so much of modern life. These professional risk takers-poker players and hedge fund managers, crypto true-believers and blue-chip art collectors-can teach us much about navigating the uncertainty of the 21st century. By embedding within the worlds of Doyle Brunson, Peter Thiel, Sam Bankman-Fried, Sam Altman, and many others, Silver offers insight into a range of issues that affect us all, from the frontiers of finance to the future of AI. The River has increasing amounts of wealth and power in our society, and understanding their mindset-including the flaws in their thinking-is key to understanding what drives technology and the global economy today. There are certain commonalities in this otherwise diverse group: high tolerance for risk; appreciation of uncertainty; affinity for numbers; skill at de-coupling; self-reliance and a distrust of the conventional wisdom. For the River, complexity is baked in, and the work is how to navigate it, without going beyond the pale. Taking us behind-the-scenes from casinos to venture capital firms to the FTX inner sanctum to meetings of the effective altruism movement, On the Edge is a deeply-reported, all-access journey into a hidden world of powerbrokers and risk takers"--
