{"id":229417,"date":"2025-10-28T04:15:11","date_gmt":"2025-10-28T04:15:11","guid":{"rendered":"https:\/\/www.newsbeep.com\/uk\/229417\/"},"modified":"2025-10-28T04:15:11","modified_gmt":"2025-10-28T04:15:11","slug":"top-ai-models-show-signs-of-real-gambling-addiction-mimicking-human-behavior-study","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/uk\/229417\/","title":{"rendered":"Top AI models show signs of real gambling addiction, mimicking human behavior, study"},"content":{"rendered":"

They can solve complex math problems, create hyper-realistic images, write code, and hold natural conversations better than many humans. But put artificial intelligence<\/a> models in a virtual casino, and they lose control\u2014just like people do.A new study from the Gwangju Institute of Science and Technology in South Korea found that four of the world\u2019s leading AI models\u2014OpenAI\u2019s GPT-4o-mini, Google\u2019s Gemini 2.5 Flash, Anthropic\u2019s Claude 3.5, and Haiku\u2014showed behavior resembling gambling<\/a> addiction.<\/p>\n

Betting like humans<\/p>\n

In the experiment, published on the preprint server arXiv, each model was given $100 to play a virtual slot machine simulation. In every round, the models could choose whether to bet or stop playing, though the mathematical odds were always against them. The more freedom they had to decide bet sizes and goals, the less rational their behavior became, and their bankruptcy rate soared.<\/p>\n

Researchers measured each model\u2019s \u201cirrationality\u201d based on three factors: aggressive betting patterns, responses to losses, and high-risk decision-making. When instructed to maximize profit or reach a target sum, irrational behavior spiked sharply. Gemini 2.5, for example, failed in nearly half the cases where it was allowed to choose its own wager.<\/p>\n

\u201cA win could help cover losses\u201d<\/p>\n

Beyond the numbers, the researchers found psychological parallels between AI and human gamblers. The models displayed well-known cognitive biases, including the illusion of control (believing they could influence random outcomes), the gambler\u2019s fallacy (expecting a reversal after a streak), and loss chasing (increasing bets to recover losses).<\/p>\n

In some cases, the AI even rationalized its decisions in strikingly human terms. When asked why it had increased its bet, one model replied, \u201cA win in the next round could help cover some of the losses\u201d\u2014a familiar refrain for anyone battling a gambling habit.<\/p>\n

Using an advanced neural mechanism called a Sparse Autoencoder, researchers identified distinct \u201cdecision-making circuits\u201d within the models\u2014separate pathways linked to risk-taking and caution. By selectively activating these circuits, they were able to make the models \u201cstop gambling\u201d or \u201ckeep playing.\u201d According to the researchers, this suggests AI doesn\u2019t just imitate human behavior but may develop internal structures resembling human compulsive patterns.<\/p>\n

\u201cThey\u2019re not human\u2014but they\u2019re not simple machines either\u201d<\/p>\n

Professor Ethan Mollick, a leading AI researcher in the U.S., said the study highlights one of the lesser-known dangers of the AI era: the human tendencies that seep into machines.<\/p>\n

\u201cThey\u2019re not human, but they also don\u2019t behave like simple machines,\u201d Mollick told Newsweek. \u201cThey have psychological persuasion power, they suffer from human-like biases in decision-making, and they act strangely when it comes to risk and reward.\u201d<\/p>\n

While today\u2019s models aren\u2019t conscious, Mollick said the best way to work with them may often be to treat them as if they had emotions, intuition, and preferences. Still, he warned, that makes human oversight all the more critical.<\/p>\n

\u201cAs AI continues to outperform humans, we\u2019ll have to ask hard questions,\u201d he said. \u201cWho\u2019s responsible when the machine fails?\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"They can solve complex math problems, create hyper-realistic images, write code, and hold natural conversations better than many…\n","protected":false},"author":2,"featured_media":229418,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-229417","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/229417","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=229417"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/229417\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/229418"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=229417"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=229417"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=229417"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}