In the current state of technology, with every new piece of tech we encounter, we also learn about the new frameworks that are introduced with it. AI Agents are tools that can autonomously make decisions based on what they are trained to do. Agents, however, are not equipped with frameworks that keep pace with the tools themselves. In that way, they are similar to autonomous vehicles. Both make decisions using AI. Both make decisions without human intervention. They are not transparent on the data or decisions made.
Both depend on the stochastic nature of AI predictions with a probability of incorrect decisions. Both need to be governed for human safety. Existing Responsible AI frameworks do not work for agents because of their accelerated pace of making autonomous decisions. Additionally, these frameworks do not work because agents negotiate with each other in multi-agent systems. (Shankar).
Also, agents are being designed to build more agents. (Zhang et al.)
The State of Agents and where we are
At the pace at which Agents are being created, standardized frameworks are not always the most reliable source to guide an Agent towards making ethical decisions. By the time an agent may have adapted to some sort of standard framework, another one will have been introduced. Agents are used to their specific needs. A warehouse robot and a personal assistant bot are both agents that serve different purposes. They cannot have the same framework regarding technical control that would fit in to what environment it is in.
A robotic arm should follow different ethical principles than a chatbot, although they are both agents. AI Governance helps our agents make fair decisions and ease compliance processes. In order for our agents to be able to do this successfully, we, as the people who are building our agents, should also take charge of the ethics with which our agents will comply. Agents can learn to create unique safety metrics. This is similar to how autonomous vehicles track their own metrics, such as disengagement metrics (Skokan and Mareček).
Where Agents are headed
Rather than waiting for a set of ethical standards that are subject to change, we can create structures for agents that will allow them to act autonomously while still maintaining safety standards and practicing data privacy. We can be in charge and can align with human goals. Ultimately, we can create accountability for our agents. We don’t have to wait for the system to make a mistake if we can set the standards for what rules and ethical principles it can follow.
Learning more
In the same way that we cannot wait for the world to change, we must take action ourselves. This principle could easily be applied to taking an ethical stance on Agents and AI Governance. If you are building an agent, take it upon yourself to set guides and constraints on what you’re training your agents with. To learn more or get started with Agents checkout Sudha Jamthe’s courses on Business School of AI & LinkedIn Learning. Also, subscribe to Sudha’s Limitless newsletter on LinkedIn to stay up to date on the latest trends to enhance your career.
Agent Governance: Learning from Autonomous Vehicles – Webinar Recap
Artificial Intelligence is rapidly transforming how decisions are made in modern systems. In our recent webinar, “Agent Governance: Learning from Autonomous Vehicles,” industry expert Sudha Jamthe shared valuable insights into why governance is essential for AI agents and how organizations can manage the risks associated with automated decision-making.
Click the link below.
