Urgent.News

What's breaking now, across thousands of outlets.

AI

Using AI for Just 10 Minutes Erodes Your Ability to Persist at Hard Things

A UC Berkeley researcher, Brian Christian, co-authored a study highlighting the potential negative effects of using artificial intelligence for just 10 minutes on one's ability to persist at difficult tasks. Christian, a research fellow at the UC Berkeley Center for Human-Compatible AI, has been studying AI's impact on our minds and the world for 20 years. His 2020 book, The Alignment Problem, has been widely acclaimed as one of the best resources on AI.

Christian and his team conducted a series of randomized controlled trials, recruiting 1,222 participants online. They divided participants into two groups: one that solved 15 basic fraction problems without AI assistance and another that used ChatGPT for help. After solving 12 problems, AI assistance was removed, and researchers observed a significant decline in accuracy and persistence in the AI-assisted group. The same pattern emerged in a larger group of 667 participants.

The researchers also tested this phenomenon with an SAT reading comprehension prompt, and the results were consistent with the previous experiments. When the AI tool was no longer available, participants' persistence and accuracy dropped significantly. Christian emphasized that this finding is a clear indication of how AI reliance can negatively impact our cognitive abilities and the value of "productive struggle" in learning.

The study, which involved scholars from various prestigious institutions, has garnered attention from major media outlets, including WIRED, Psychology Today, and The New York Times. Christian argues that AI should be used as a tool to augment human abilities rather than replace them, especially in research and academia. He worries that overreliance on AI may undermine essential qualities in researchers, such as close data engagement and teamwork.

The study highlights the potential risks and opportunities for human rights fields as AI continues to advance.

Written by urgent.news from Hacker News's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at vcresearch.berkeley.edu →

More in AI

The Gateway, the Registry and the Executor: Keeping Groq at the Edge of My Backend

SokoFlow AI Build Log, Week 2 of 6: the LLMClient, the ToolExecutor, and one component that almost became a god object The Week 2 goal said "wire three tools end-to-end." By the end of the week, all…

  • LLMClient acts as gateway, translating requests between SokoFlow and providers
  • ToolExecutor handles tool execution, runs against PostgreSQL database
  • ToolRegistry provides centralized tool management for scalability

Touch Grass AI

🌱 Touch Grass AI: An Offline Garden Planner Powered by Open-Weight Models This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass What I Built Touch Grass AI is an…

More from Sunday 11 October →