{"id":18198,"date":"2024-10-30T13:50:15","date_gmt":"2024-10-30T11:50:15","guid":{"rendered":"https:\/\/forklog.com\/en\/scientists-develop-ai-model-mimicking-human-behavior\/"},"modified":"2024-10-30T13:50:15","modified_gmt":"2024-10-30T11:50:15","slug":"scientists-develop-ai-model-mimicking-human-behavior","status":"publish","type":"post","link":"https:\/\/forklog.com\/en\/scientists-develop-ai-model-mimicking-human-behavior\/","title":{"rendered":"Scientists Develop AI Model Mimicking Human Behavior"},"content":{"rendered":"

A team of experts from several universities has developed a computational model that predicts and mimics human behavior with high accuracy, according to a research paper<\/a>.<\/p>\n

Centaur is trained on a dataset called Psych101, which offers an “unprecedented scale of information on human behavior.” The model is “capable of accelerating experiments for scientific discoveries.”<\/p>\n

Psych101 encompasses 160 psychological experiments involving 60,092 individuals, who made a total of over 10 million choices.<\/p>\n

\n

\u201cPrevious attempts have been made to embody such theories by building computational models, yet we currently lack one that fully reflects the human mind,\u201d the researchers noted.<\/p>\n<\/blockquote>\n

\"Scientists
Scientists claim the new model can mimic and predict human behavior in any field. Data: Centaur study: a foundation model of human cognition.<\/figcaption><\/figure>\n

Today, most modern computational models are tailored to specific domains. They are designed to solve particular problems, the experts noted.<\/p>\n

\n

\u201cThe human mind is remarkably versatile. We not only regularly make mundane decisions like choosing breakfast cereal or attire, but also tackle complex challenges such as finding ways to cure cancer or explore space,\u201d they added.<\/p>\n<\/blockquote>\n

Lead scientist Marcel Binz noted that the team refined the modern LLM<\/span> Llama 3.1 70B by Meta<\/a> and discovered its \u201cability to predict the behavior of unseen participants better than existing cognitive models in almost every experiment.\u201d<\/p>\n

\n

We then finetuned a state-of-the-art language model (Llama 3.1 70B) on this data set and found that the resulting model predicts behavior of unseen participants better than existing cognitive models in almost every single experiment. pic.twitter.com\/AS3BVutlpK<\/a><\/p>\n

\u2014 Marcel Binz (@marcel_binz) October 28, 2024<\/a><\/p><\/blockquote>\n