# New research published in the journal Computers in Human Behavior suggests that using AI tools like ChatGPT to complete tasks can improve a person’s performance while simultaneously distorting their… **By:** Jonathan Boymal **Published:** 2025-12-30T06:12:55.899Z **Source:** [LinkedIn](https://www.linkedin.com/posts/jonathan-boymal-%F0%9F%90%9D-448b5870_new-research-published-in-the-journal-computers-activity-7411650478683422721-ihff?rcm=ACoAAAfsct4BVBNkOVI5cyYn-UOIIe2RSRhT-cI) --- # Jonathan Boymal’s Post [Jonathan Boymal](https://au.linkedin.com/in/jonathan-boymal-448b5870?trk=public_post_feed-actor-name) New research published in the journal Computers in Human Behavior suggests that using AI tools like ChatGPT to complete tasks can improve a person’s performance while simultaneously distorting their ability to assess that performance accurately. The research team designed two distinct studies centered on logical reasoning tasks. In the first study, they recruited 246 participants from the United States. These individuals were asked to complete 20 logical reasoning problems taken from the Law School Admission Test (LSAT). The researchers provided participants with a specialised web interface. This interface displayed the questions on one side and a ChatGPT interaction window on the other. Participants were required to interact with ChatGPT at least once for each question. They could ask ChatGPT to solve the problem or explain the logic. After submitting their answers, participants estimated how many of the 20 questions they believed they had answered correctly. They also rated their confidence on a specific scale for each individual decision. The results of this first study showed a clear improvement in objective performance. On average, participants using ChatGPT scored approximately three points higher than a historical control group of people who took the same test without AI assistance. Despite this improvement in scores, the participants engaged in significant overestimation. On average, the group estimated they had answered about 17 out of 20 questions correctly. In reality, their average score was closer to 13. A significant theoretical contribution of this research involves the Dunning-Kruger effect. In typical scenarios without AI, the data would show a steep slope where low performers vastly overestimate themselves and high performers do not. When participants used ChatGPT, this effect vanished. The “leveling” effect of the technology meant that overestimation became uniform across the board. Low performers and high performers alike inflated their scores by similar amounts. To ensure these findings were robust, the researchers conducted a second study. This replication involved 452 participants. The researchers split this sample into two distinct groups. One group performed the task with AI assistance, while the other group worked without any technological aid. In this second experiment, participants were told they would receive a financial bonus if their estimate of their score matched their actual score. The results of the second study mirrored the first. The unaided group showed the classic Dunning-Kruger pattern, where the least skilled participants showed the most bias. The AI group again showed a uniform bias, confirming that the technology fundamentally shifts how users perceive their competence. The researchers discussed several potential reasons for these outcomes, including the “illusion of explanatory depth.” This is a beautiful empirical snapshot of something, that from an ontocybernetic perspective, has been absehbar for years: AI upgrades performance, while downgrading orientation. What these experiments really show is less about reasoning and more about meta-reasoning: the tool raises the floor of task results, but it also flattens the landscape of self-assessment. The classical Dunning–Kruger curve verschwindet nicht, sie wird technologisch „nivelliert“ – wir erhalten eine gleichmäßig verteilte Illusion von Kompetenz. In meinen Arbeiten nenne ich das eine Störung der epistemic health: the distinction between doing something with AI and knowing what one is actually doing erodes. The system becomes collectively smarter on paper, while subject-level judgement becomes opaque to itself. The real design question is therefore: How do we build an Epistemic Core / Orientation Layer in which AI assistance is framed, so that every gain in output is accompanied by a gain in calibration – in the ability to see where and why one might still be wrong? Genau an dieser Stelle setzt das Epistemic Integrity Umbrella an – mehr dazu im Framework hier: 👉 [https://leontsvasmansapiognosis.substack.com](https://leontsvasmansapiognosis.substack.com?trk=public_post_comment-text) This research doesn’t show that AI makes people worse thinkers. It shows what happens when performance improves faster than authorship clarity. When humans don’t carry their side of the reasoning, metacognition collapses. When they do, AI becomes a precision tool — not a confidence amplifier.