Source Details

{
    "source_type": "extension",
    "content_hash": "e618ffd9011dabe774f62fc97a89607c63fc378bb3ebf8b92f00778e8de23343",
    "submitted_via": "chrome_extension",
    "extension_version": "1.0.18",
    "original_url": "https:\/\/www.nytimes.com\/interactive\/2026\/09\/29\/magazine\/ai-chatbots-brain-development-study.html?emc=edit_na_20260930&nl=breaking-news&segment_id=227304",
    "parsed_content": "Machine Learnings Will A.I. Make Your Brain Lazy? Here\u2019s What the New Research Actually Shows. Share full article Illustrations by Juliana Castro Var\u00f3n By Juliana Castro Var\u00f3n and Dylan Freedman Sept. 29, 2026 Machine Learnings is a column examining what the latest scientific studies reveal about our increasingly complex relationship with artificial intelligence.Artificial intelligence is changing the world, and the humans in it, in ways we don\u2019t yet understand. People are increasingly worried that if we rely on chatbots for thinking, our brains will become slower and less engaged. Is A.I. making us stupider? Scientists, too, want to know.A team of researchers from the University of California, Irvine, and the education company McGraw Hill examined a decade of data from a popular online learning tool by McGraw Hill, starting in 2015. Across millions of math sessions by a group that ranged from fifth graders to college students, the researchers compared performance from before 2022, when ChatGPT first came out, with the work that followed its arrival. To analyze the results, they split the questions into two types. First: text-based problems that students could easily copy into a chatbot.And next: visual, interactive or multistep questions that were harder for a chatbot.Researchers found that after 2022, the time students spent mulling over online test questions dropped precipitously on the questions readily handled by chatbots, while their performance improved.But when the students were given proctored tests away from their computers, their results were worse than they had been before A.I. Sina Rismanchian, a Ph.D. candidate at U.C. Irvine and one of the lead authors of the preliminary study, noted that A.I. lets students offload the process required to solve a mathematical problem.Another multi-institutional study, prepared for the forthcoming Conference on Language Modeling and cited by Rismanchian and his co-authors, shows that when people (in this experiment, adults) become accustomed to skipping the productive struggle of logical and analytical tasks and having the process automated, they are more likely to give up on hard questions that they could otherwise figure out.A.I. tools condition people to expect immediate responses and deny them the experience of working through challenges, the authors say, which is crucial for skill development and one of the strongest predictors of long-term learning.\u201cPart of how A.I. is making us dumber,\u201d says Dr. Adam Green, a cognitive neuroscientist at Georgetown University, is \u201cundercutting the development of learning how to think in younger people that now have an alternative.\u201dGreen compares the brain\u2019s tendency to conserve effort to calorie retention.\u201cWhatever you could do that exerted the least calories but didn\u2019t kill you was the best thing to do evolutionarily,\u201d he says.But all is not lost. Studies suggest that \u2014 for the brain \u2014 there are better ways to use A.I. SKIP ADVERTISEMENT In July, professors at Middlebury College released a preprint study in which more than 200 undergraduates were asked to write an essay and take a proctored multiple-choice test.One group of students had access to A.I. to research a topic and write the essay. The other group didn\u2019t. The researchers found that the A.I.-enabled group wrote longer and slightly higher-quality essays. Their test scores were also higher.A week later, the participants returned for a similar test, but now both groups worked unassisted. The authors broke down the students\u2019 second-round performance based on how A.I. was used in the first round. The students who relied on A.I. to draft their entire essays in the first round floundered. Their test scores decreased.But the students who had used A.I. more like a tutor, to explain complex concepts or clarify ideas, not only retained high test scores in the second round, but their essays were higher quality as well. Experimental evidence suggests that using A.I. not as a full shortcut, but as a means to exercise what they were supposed to learn, allowed them to outperform the remaining participants in the A.I. group and those who hadn\u2019t used A.I. at all.In other words, these studies so far agree: Artificial intelligence is affecting how people, especially the young, learn and remember. Whether we will be dumber as a result might depend on how \u2014 not if \u2014 we use A.I. Juliana Castro Var\u00f3n is the senior design editor of A.I. Initiatives at The Times. She produces interactive stories about technology and culture, and designs software for investigative reporting. Dylan Freedman is the A.I. projects editor for The Times. He reports on artificial intelligence and uses computational tools to unlock stories from vast troves of data.The studies:\u201cFaster Completion, Less Learning: Generative A.I. Reduced Study Time on Math Problems and the Knowledge They Build\u201d (2026).\u201cA.I. Assistance Reduces Persistence and Hurts Independent Performance\u201d (2026).\u201cExperimental Evidence on the Learning Impact of Generative A.I.\u201d (2026)",
    "published_at": "2026-09-29T20:34:51+00:00",
    "author_name": "Juliana Castro Var\u00f3n, Dylan Freedman",
    "site_name": "nytimes.com",
    "section": "Machine Learnings",
    "publisher": "The New York Times",
    "ai_headline": "Will A.I. Make Your Brain Lazy? Here\u2019s What the New Research Actually Shows.",
    "ai_simplified_title": "New Research Shows AI Impact on Brain Development",
    "ai_excerpt": "New research suggests that while AI can improve performance on certain tasks, over-reliance on it can hinder critical thinking and long-term learning. The studies indicate that how AI is used, either as a shortcut or a learning aid, determines its impact on cognitive development.",
    "ai_subject_tags": [
        "Artificial Intelligence",
        "Education",
        "Cognitive Science",
        "Learning",
        "Brain Development",
        "Research"
    ],
    "ai_context_type": "Analysis",
    "ai_context_details": {
        "tone": "informative",
        "perspective": "academic",
        "audience": "general",
        "credibility_indicators": [
            "expert_quotes",
            "data_cited",
            "research_studies"
        ]
    },
    "ai_source_vector": null,
    "ai_confidence_score": 0.9999999999999999,
    "ai_extraction_metadata": {
        "extracted_at": "2026-09-30T18:01:26.455673Z",
        "ai_model": "gemini-2.5-flash",
        "extraction_method": "automated",
        "content_length": 5084,
        "metadata_prompt_body_chars": 5084,
        "metadata_prompt_truncated": false,
        "url": "https:\/\/nytimes.com\/interactive\/2026\/09\/29\/magazine\/ai-chatbots-brain-development-study.html",
        "existing_metadata": {
            "author_name": null,
            "published_at": null,
            "domain_name": "nytimes.com",
            "site_name": null,
            "section": null,
            "publisher": null
        },
        "openai_refusal_gemini_fallback": false
    }
}

{
    "extracted_at": "2026-09-30T18:01:26.455673Z",
    "ai_model": "gemini-2.5-flash",
    "extraction_method": "automated",
    "content_length": 5084,
    "metadata_prompt_body_chars": 5084,
    "metadata_prompt_truncated": false,
    "url": "https:\/\/nytimes.com\/interactive\/2026\/09\/29\/magazine\/ai-chatbots-brain-development-study.html",
    "existing_metadata": {
        "author_name": null,
        "published_at": null,
        "domain_name": "nytimes.com",
        "site_name": null,
        "section": null,
        "publisher": null
    },
    "openai_refusal_gemini_fallback": false
}