Artificial intelligence (AI) is all the rage nowadays. This is largely because of advances in large language models (LLMs) that can chat on just about any topic. Although it is interesting to be able to chat to these artificial minds, about this and that, a race has now begun to turn general purpose chatbots into ‘Agents’ that can carry out specific tasks.
If that is successful, we may soon find that LLM powered Agents begin to take over from humans in many different fields and occupations. But before anything like that happens, we really should try to understand how to design these Agents so that they work well with human psychology, playing to our strengths and avoiding our weaknesses.
In this online talk on Thursday 22 October, from 4pm-5pm (BST), Dr Kevin Allan, will describe how some major concerns about LLMs acquiring and expressing social stereotypes, e.g. about gender or age, could amplify existing stereotypical biases in our decision-making, and make real-world inequalities significantly worse.
Dr Allan will show you how we are trying to safely control and investigate such phenomena in University labs, so that we can apply that understanding to real-life human-AI interactions and develop novel technologies to detect and counteract such influences.
If you would like to attend this event please register below and joining information will be shared with you closer to the time.
- Speaker
- Contact
- Add to Calendar