Experts Give AI "Motivation And Memory", Put Them In Simulated Town. Results Are StunningArtificial intelligence (AI) has been making rapid strides, learning new things and evolving into a smarter system. Now, a Wharton professor has posted a couple of tweets detailing how AI-powered characters engaged in complex behaviour when given "motivations and memory" and put in a simulated town. Professor Ethan Mollick said in his tweets that their actions were rated "more human than humans roleplaying". He also posted link to a study that gave details about the experiment, including a computer game-like simulation where users can actually see the "AI agents" at work.
"This is quite the paper! It gave 25 AI agents motivations & memory, and put them in a simulated town. Not only did they engage in complex behavior (including throwing a Valentine's Day party) but the actions were rated more human than humans roleplaying," Professor Mollick said in his tweet.
The accompanying simulation shows AI generative agents simulating day-to-day human behaviour, such as waking up, cooking breakfast and heading to work.
The agents tagged as artists paint, while authors write. They are also seen forming opinions, noticing each other and initiating conversations. They also remember and reflect on days past and plan for the next day - a major memory feat highlighted by Professor Mollick.
The simulated town appears to be based on 'The Sims'. The study said these generative agents produced believable individual and emergent social behaviours.
Experts have always raised concerns around developing increasingly smarter AI models. And these have exacerbated after the launch ChatGPT, a chatbot that engages in conversation with humans and produces remarkable results.
Many people - from students to politicians - have made use of ChatGPT's capabilities in writing speeches or exams and solving complex mathematical problems.
However, more than 1,000 technology leaders and researchers, including Twitter chief Elon Musk, have urged artificial intelligence labs to pause development of the most advanced systems, warning in an open letter that AI tools present "profound risks to society and humanity".
"This is quite the paper! It gave 25 AI agents motivations & memory, and put them in a simulated town. Not only did they engage in complex behavior (including throwing a Valentine's Day party) but the actions were rated more human than humans roleplaying," Professor Mollick said in his tweet.
The accompanying simulation shows AI generative agents simulating day-to-day human behaviour, such as waking up, cooking breakfast and heading to work.
The agents tagged as artists paint, while authors write. They are also seen forming opinions, noticing each other and initiating conversations. They also remember and reflect on days past and plan for the next day - a major memory feat highlighted by Professor Mollick.
The simulated town appears to be based on 'The Sims'. The study said these generative agents produced believable individual and emergent social behaviours.
Experts have always raised concerns around developing increasingly smarter AI models. And these have exacerbated after the launch ChatGPT, a chatbot that engages in conversation with humans and produces remarkable results.
Many people - from students to politicians - have made use of ChatGPT's capabilities in writing speeches or exams and solving complex mathematical problems.
However, more than 1,000 technology leaders and researchers, including Twitter chief Elon Musk, have urged artificial intelligence labs to pause development of the most advanced systems, warning in an open letter that AI tools present "profound risks to society and humanity".