Introducing the Artificial Agency Speaker Series
Click here to join our mailing list and hear about future events!
UK AI Forum is proud to announce its inaugural speaker series, running through October-December 2025. This series will explore the theme of Artificial Agency from multiple perspectives.
The first mass commercial systems in the recent resurgence of AI were chatbots, with ChatGPT chief among them. Although impressive, these systems had several limitations which hindered their use in real-world contexts. For instance:
they were unable to effect change in the real world, beyond the confines of their chat interface;
they could not act autonomously without user input
they could not coherently pursue goals over an extended time-frame
they were unable to learn and accumulate knowledge which carried over between interactions.
In short, these systems lacked agency. Since 2023, significant advancements have been made in ameliorating these problems. Present day agentic systems are proving effective across a wide range of contexts, and progress is continuing at a heady pace. Our speaker series will examine the recent technical progress in artificial agents, the anticipated societal impacts of widespread use of such agents, and the conceptual frameworks we should use to reason about these systems.
Technical advances in Artificial Agency
The time-horizon over which AI agents can act successfully is increasing at an exponential rate. What future progress can we expect from artificial agents in the next year? When will such agents be able to autonomously replicate themselves? We’ll explore what technical advances underlie the success of recent progress, and whether scaling really is all you need. We’ll also be asking what technical challenges need to be solved to allow for the safe deployment of powerful agentic systems. For example, how can artificial agents be controlled, and what safeguards are necessary to prevent such agents from taking harmful actions?
Societal Impacts of Artificial Agency
As artificial agents become ever more capable and coherent, so does the range of situations they can be expected to be deployed in. How can we best prepare for a world populated by large quantities of artificial agents? Within this subtopic, we’ll be asking what economic ramifications of cheap, competent, and widely available AI agents are; what regulatory and legal frameworks are necessary for the widespread deployment of such agents; and what the psychological and interpersonal consequences of interacting far more with artificial agents might be.
Conceptual foundations of Artificial Agency
How should we conceptualise and think about artificial agents? We’ll investigate how artificial agency differs from natural, biological agency, and how agency differs between biological systems. We’ll think about the dynamics of interactions between agents, and what novel dynamics might emerge when agents interact with other instances of themselves. We’ll ask about the way artificial agents conceptualise themselves, and what goals powerful artificial agents might be expected to pursue when allowed to freely operate in the world.