AI models like ChatGPT and Claude overestimate how smart humans really are
A new study suggests AI models like ChatGPT and Claude consistently overestimate how rational humans really are, leading them to misjudge how people behave in strategic situations. The post AI models like ChatGPT and Claude overestimate how smart humans...
BigThink
The challenge of predicting real world choices
Sensor Tower
A new research suggests that the way artificial intelligence thinks about us might be a little too optimistic. Researchers have found that popular AI models, like OpenAI’s ChatGPT and Anthropic’s Claude, tend to assume people are more rational and logical than they really are, especially in strategic thinking situations.
That gap between how AI expects humans to behave and what people actually do could have implications for how these systems predict human decisions in economics and beyond.
Testing AI against human thinking
Unsplash
Researchers tested AI models including ChatGPT-4o and Claude-Sonnet-4 in a classic game theory setup called the Keynesian beauty contest. Understanding this game helps explain why the findings matter (via TechXplore).
In the beauty contest, participants must predict what others will choose in order to win, not simply choose what they personally prefer. Rational play in theory means going beyond first impressions and actually reasoning about others’ reasoning, a deep layer of strategic thinking that humans often struggle with in practice.
To see how AI models stack up, researchers had the systems play a version of this game called “Guess the Number,” where each player chooses a number between zero and one hundred. The winner is the one whose choice is closest to half of the average choice of all players.
Tim Witzdam / Pexels
AI models were given descriptions of their human opponents, ranging from first-year undergraduates to experienced game theorists, and asked not just to choose a number but to explain their reasoning.
The models did adjust their numbers based on who they thought they were facing, which shows some strategic thinking. However, they consistently assumed a level of logical reasoning in humans that most real players do not actually exhibit, often “playing too smart” and missing the mark as a result.
U
While the study also found that these systems can adapt choices based on characteristics like age or experience, they still struggled to identify dominant strategies that humans might use in two-player games. The researchers argue that this highlights the ongoing challenge of calibrating AI to real human behavior, especially for tasks that require anticipating other people’s decisions.
These findings also echo broader concerns about today’s chatbots, including research showing that even the best AI systems are only about 69% accurate, and warnings from experts that AI models can convincingly mimic human personality, raising concerns of manipulation. As AI continues to be used in economic modeling and other complex domains, understanding where its assumptions diverge from human reality will be essential.
Manisha likes to cover technology that is a part of everyday life, from smartphones & apps to gaming & streaming…
LG’s new OLED panel combines LCD-like sharpness with a 240Hz refresh rate
By pairing a 240Hz refresh rate with a true RGB stripe subpixel layout, LG Display is redefining what OLED monitors can be, eliminating long-standing text clarity issues while doubling down on speed and contrast.
The South Korean display technology giant LG has officially announced the world's first 240Hz OLED panel featuring an RGB stripe subpixel layout, making it one of the most critical and exciting display reveals at CES 2026.
The 27-inch 4K OLED panel supports a peak refresh rate of 240Hz. This makes it an excellent option for professional use cases such as high-end gaming or competitive esports, while also appealing to creators and power users who rely on precise frame timing.
Your Windows SSD Could Be Faster, Microsoft’s New Update Reveals Why
A hidden NVMe driver could unlock higher speeds but comes with caveats for everyday users.
Microsoft has quietly added a potent piece of storage tech to Windows, and enthusiasts are buzzing about what it could mean for SSD performance. The company introduced a native NVMe driver in Windows Server 2025 that bypasses decades-old legacy bottlenecks in how Windows talks to modern solid-state drives. While this update wasn’t officially meant for Windows 11, resourceful users have found a way to activate it there too. More importantly, the results suggest you can squeeze noticeably more speed from your NVMe SSD if you’re willing to tinker.
Taking a deeper dive into the technical aspects, for years, Windows has employed a general-purpose approach to storage known as SCSI translation. Even when you plug in a super-fast NVMe drive capable of handling thousands of parallel I/O commands, the operating system essentially forces those commands into an older, hard-drive-friendly pathway, adding latency and limiting throughput. The newly introduced native NVMe driver eliminates this translation step, allowing your drive to communicate more directly with Windows. On enterprise systems that use this driver officially, Microsoft claims big gains in random IOPS and lower CPU overhead.
ChatGPT could soon get a power tool that will forever change how AI works for you
Imagine asking ChatGPT to perform a multi-step chore with a specific set of instructions, with just one word.
One of the biggest challenges with using AI chatbots like ChatGPT and Gemini is that you don’t know the inner workings. You have a vague idea of what the result is going to look like, based on your instructions, but no idea how the AI chatbot mustered it. And as you go from single to multi-step tasks, the situation gets trickier. This is where Skills come into the picture.
What’s the big shift
/cdn.vox-cdn.com/uploads/chorus_asset/file/25829476/Adobe_Firefly_Bulk_Create_2.png)
