Age of Agents

“What’s an agent?”, my friend asked. To me it’s quite obvious from the word itself. Agent is an entity which has “agencies” - power or capacity to fulfil an action or a service. And agents act on behalf of someone. Today, we want AI models to be the agents - agents to perform actions on our desktop or through the internet of things.

How does it work? The AI models, acting as agents, need knowledge and tools. The knowledge to answer questions, make decisions, or to learn how to use tools. Tools to do things, make changes, send emails, book tickets, and in course of time, actions of much more significant consequences.

The AI models, acting as agents, need knowledge and tools. The knowledge to answer questions, make decisions, or to learn how to use tools.

Where do we stand now? Today our knowledge is accumulated and managed for our consumption, mostly for humans. Though the large language models imitate human-like reasoning, without conceptual and physical models it may not be able to justify its actions well enough. Over time, the models should be able to work with abstract concepts and decisions and then translate it to a language. The AI agents will possess both skills & resources (abstract & concrete knowledge).

How about the tools then? There would be two kinds of tools, in my opinion. Digital tools (programmatic actions) and physical tools (robotic actions), both interlinked in many different ways. In the short term, it would be easier to get the software tools working. Invoke APIs, automate business processes, replace ERP software, and control internet-enabled toasters.

So, will there be an AI revolution soon? Wait! Before we give the keys to AI, we have to ensure the reliability aspects of AI Agents. The knowledge repositories and tools of today are not built for AI. We are stretching the AI to mimic people and systems to be able to use those.

You should read about how we transitioned from horse-drawn carriages to horseless carriages (cars). During the early, days the cars were called “devil wagons”; they came without adequate braking power and other issues with reliability and incompatibilities with the environment. Newspapers both hailed and criticised the new technology. People both hated and loved it. Eventually, we adapted, improved it, made it reliable, changed the roads, systems, and rules, and now we have made it autonomous (well, almost). In our minds AI Agents are similar to those autonomous cars, but in fact they are closer to those first generation cars.