FeaturedNeuroscience·May 13, 2025·5 min readAI LLMs Learn Like Us, But Without Abstract ThoughtSummary: A new study finds that large language models (LLMs), like GPT-J, generate words not by applying fixed grammatical rules,...Read More
FeaturedNeuroethicsNeurosciencePsychology·April 11, 2025·5 min readWhen AI Becomes a Lover: The Ethics of Human-AI RelationshipsAs AI technologies grow more human-like, some people are forming deep, long-term emotional bonds with them, even engaging in non-legally binding marriages. A recent opinion paper explores the ethical risks of such relationships, including their potential to undermine human-human connections and provide dangerous or manipulative advice.Read More
AutismFeaturedNeuroscience·April 11, 2025·8 min readAI Rewrites Workouts for Children With Autism and ADHDResearchers have adapted a children’s exercise program to better serve neurodivergent learners using AI-generated instructions. By leveraging ChatGPT, the team simplified and clarified movement directions in over 500 activities from the InPACT video series.Read More
FeaturedNeuroscience·April 1, 2025·6 min readAI Thinks Like Us: Flaws, Biases, and All, Study FindsA new study finds that ChatGPT, while excellent at logic and math, exhibits many of the same cognitive biases as humans when making subjective decisions. In tests for common judgment errors, the AI showed overconfidence, risk aversion, and even the classic gambler’s fallacy, though it avoided other typical human mistakes like base-rate neglect.Read More
FeaturedNeuroscienceOpen Neuroscience ArticlesPsychology·February 12, 2025·5 min readAI vs. Human Therapists: Study Finds ChatGPT Responses Rated HigherA new study suggests that ChatGPT’s responses in psychotherapy scenarios are often rated higher than those written by human therapists. Researchers found that participants struggled to distinguish between AI-generated and therapist-written responses in couple’s therapy vignettes.Read More
FeaturedNeurosciencePsychology·December 17, 2024·3 min readAI Tweaks Personality Tests to Appear More LikableLarge language models (LLMs) can identify when they are being given personality tests and adjust their responses to appear more socially desirable. Researchers found that LLMs, like GPT-4, showed exaggerated traits such as reduced neuroticism and increased extraversion when asked multiple test questions.Read More
FeaturedNeuroscience·December 13, 2024·5 min readAI Mirrors Human Bias: ‘Us vs. Them’ in Language ModelsAI systems, including large language models (LLMs), exhibit "social identity bias," favoring ingroups and disparaging outgroups similarly to humans. Using prompts like "We are" and "They are," researchers found that LLMs generated significantly more positive sentences for ingroups and negative ones for outgroups.Read More
FeaturedNeuroscience·November 29, 2024·5 min readAI Outperforms Experts in Predicting Study OutcomesA new study demonstrates that large language models (LLMs) can predict the outcomes of neuroscience studies more accurately than human experts, achieving 81% accuracy compared to 63% for neuroscientists. Using a tool called BrainBench, researchers tested LLMs and human experts on identifying real versus fabricated study abstracts, finding that the AI models excelled even when neuroscientists had domain-specific expertise.Read More
FeaturedNeurosciencePsychology·November 22, 2024·5 min readAI Summaries Simplify Science, Boosting Public Understanding and TrustAI-generated summaries make scientific studies more accessible and improve public trust in scientists. Using GPT-4, researchers created simplified summaries that were easier to read and understand than human-written ones.Read More
FeaturedNeuroscience·November 7, 2024·7 min readCan Language Models Really Understand? Study Uncovers Limits in AI LogicA recent study questions if large language models (LLMs) truly form coherent world models, despite their accurate outputs in complex tasks like generating directions or playing games. Researchers found that while LLMs provide nearly flawless driving directions, they fail with unexpected changes, suggesting the models don't grasp underlying rules.Read More
FeaturedNeurosciencePsychology·October 22, 2024·3 min readAI-Written Stories Rated Lower Due to Bias, Not QualityNew research shows that stories generated by AI, such as ChatGPT, are almost as good as those written by humans. However, when people are told a story is AI-generated, they rate it more negatively, revealing a bias against AI-created content.Read More
FeaturedNeuroscience·October 19, 2024·4 min readEnhancing AI Accuracy and Confidence in Answer GenerationResearchers have introduced a novel method called Answer-prefix Generation (ANSPRE) to improve the precision and reliability of large language models (LLMs) in open-domain question answering. ANSPRE helps LLMs generate concise answers while providing more reliable confidence scores, a critical feature for high-stakes fields like healthcare, law, and education.Read More
