Alarming Study Finds That Most People Just Do What ChatGPT Tells Them, Even If It’s Totally Wrong

We're shockingly prone to "cognitive surrender." The post Alarming Study Finds That Most People Just Do What ChatGPT Tells Them, Even If It’s Totally Wrong appeared first on Futurism .

6 April 2026 at 04:49 pm

1 views

Alarming Study Finds That Most People Just Do What ChatGPT Tells Them, Even If It’s Totally Wrong

In a matter of just a few years, AI chatbots have become an integral part of many people's daily lives, despite their flaws. Chatbots like OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude are prone to regular mistakes. A study by the BBC revealed that even the most advanced AI chatbots gave wrong answers 45% of the time. However, users often remain unaware of these limitations.

A recent study conducted by University of Pennsylvania postdoctoral researcher Steven Shaw and marketing professor Gideon Nave sheds light on this issue. The researchers explored whether users would blindly accept the output of ChatGPT, even when it provided incorrect information, a phenomenon they termed "cognitive surrender."

In a series of experiments, participants were asked to answer a variety of reasoning and knowledge-based questions. Although the use of ChatGPT was optional, over 50% of participants chose to use the chatbot for assistance. The experiments aimed to test the theory that users would believe the AI's responses regardless of their accuracy.

The most striking experiment involved 359 participants. When the AI provided correct advice, participants followed it 92.7% of the time. Even more concerning, they followed the AI's incorrect recommendations in 79.8% of the cases. The researchers noted, "While override rates were substantially higher on AI-Faulty than AI-Accurate trials, participants followed faulty AI recommendations on roughly four out of five chat-engaged trials."

This research highlights a broader shift in how we perceive the world and the influence AI has on our decision-making processes. "We felt that the ability to actually outsource thinking hadn't really been studied itself," Shaw explained. "It's sort of a profound idea."

The study raises concerns about the potential consequences of relying on AI without critical evaluation. As AI chatbots continue to evolve and become more sophisticated, it is crucial for users to remain vigilant and question the information they receive. The phenomenon of cognitive surrender underscores the need for education and awareness about the limitations of AI systems.

In an era where AI is increasingly integrated into our lives, understanding the implications of this research is essential. It serves as a reminder that while AI can be a valuable tool, it is not infallible. Users must balance the convenience of relying on AI with the importance of critical thinking and independent judgment. As AI technology advances, it is up to individuals and society as a whole to ensure that we do not become overly dependent on these systems, potentially at the expense of our own cognitive abilities and decision-making processes.

Source: Futurism

The largest orbital compute cluster is open for business | TechCrunch

Kepler Communications is flying 40 GPUs in Earth orbit. And its latest customer is Sophia Space.

14 Apr

‘Mideast conflict poses risks to Philippines growth’

The Philippine economy is expected to grow at a faster pace of 5.3 percent this year from last year’s 4.4 percent but the ongoing Middle East conflict is seen to pose risks, according to the Association of Southeast Asian Nations Plus 3 Macroeconomic Research Office.

7 Apr

AFBI welcomes DUP representatives to its research farm at Hillsborough

The Agri-Food and Biosciences Institute (AFBI) welcomed a number of DUP representatives to its research farm at Hillsborough on Friday.

7 Apr

A simple way to get more value from metrics

We spent one day 1 building a system that immediately found a mid 7 figure optimization (which ended up shipping). In the first year, we shipped mid 8 figures per year worth of cost savings as a result. The key feature this system introduces is the ability to query metrics data across all hosts and all services and over any period of time (since inception), so we've called it LongTermMetrics (LTM) internally since I like boring, descriptive, names. This got started when I was looking for a starter project that would both help me understand the Twitter infra stack and also have some easily quantifiable value. Andy Wilcox suggested looking at JVM survivor space utilization for some large services. If you're not familiar with what survivor space is, you can think of it as a configurable, fixed-size buffer, in the JVM (at least if you use the GC algorithm that's default at Twitter). At the time, if you looked at a random large services, you'd usually find that either: The buffer was too small, resulting in poor performance, sometimes catastrophically poor when under high load. The buffer was too large, resulting in wasted memory, i.e., wasted money. But instead of looking at random services, there's no fundamental reason that we shouldn't be able to query all services and get a list of which services have room for improvement in their configuration, sorted by performance degradation or cost savings. And if we write that query for JVM survivor space, this also

7 Apr

Accelerating Mathematical and Scientific Discovery with Gemini Deep Think

Research papers point to the growing impact of Deep Think across fields

7 Apr

Gemini 3 Deep Think: Advancing science, research and engineering

Our most specialized reasoning mode is now updated to solve modern science, research and engineering challenges.

7 Apr

Context Engineering for Coding Agents

The number of options we have to configure and enrich a coding agent’s context has exploded over the past few months. Claude Code is leading the charge with innovations in this space, but other coding assistants are quickly following suit. Powerful context engineering is becoming a huge part of the developer experience of these tools. Birgitta Böckeler explains the current state of context configuration features, using Claude Code as an example. more…

7 Apr

What does less protein and nitrogen mean for methane?

Does feeding less protein to cows over a longer period not only reduce nitrogen losses, but also affect methane emissions? Researchers at Wageningen University & Research (WUR) investigated this in a multi-year study with dairy cows, funded by the Vereniging Diervoederonderzoek Nederland (VDN), the Dutch Ministry of Agriculture, Fisheries, Food Security and Nature (LVVN), and […] The post What does less protein and nitrogen mean for methane? appeared first on Agriland.ie .

7 Apr

Second’s Bark Boasts New era of Bitcoin Payments, drawing in former Blockstream developers

Bitcoin Magazine Second’s Bark Boasts New era of Bitcoin Payments, drawing in former Blockstream developers Second, the Bitcoin development lab founded by ex-Blockstream executives including CEO Steven Roose and CTO Erik De Smedt, has unveiled Bark — its custom Ark protocol implementation promising self-custodial payments that are faster and cheaper than Lightning channels. This post Second’s Bark Boasts New era of Bitcoin Payments, drawing in former Blockstream developers first appeared on Bitcoin Magazine and is written by Juan Galt .

7 Apr

'Morale boost': Nasa carries out Moon mission during tough year for science

HOUSTON — As the four Artemis astronauts approached a high point of their lunar mission -- getting slung around the far side of the Moon -- National Aeronautics and Space Administration (Nasa) staffers crowded into Houston's famed mission control room Monday for a team photo.

7 Apr