AI’s Coming of Agents (Part 2 of 10)

In the last article, I highlighted that the 5th Level of AI is Autonomy.

Autonomy and agency are somewhat synonymous. Autonomy refers to the freedom to make choices, while agency is the ability to act on those choices. This forms the foundation for the term AI Agents which are simply programs that act as proxies on your behalf.

Let's begin by explaining, in simple terms what an AI agent is, followed by an overview of the different types of agents and a description along with basic applications.

An AI agent is like a computer program that can perceive, think, and act on its own. It's designed to perform tasks or solve problems in ways that mimic human intelligence (i.e., what you would do in a given situation).

There are different types of AI agents based on their capabilities. Here are the most common ones:

1. Reactive Agents: These agents respond to the current situation without considering past experiences or future consequences, like a simple reflex.

Example: A thermostat.

How it works: When the temperature falls below a set point, the thermostat turns on the heater. When the temperature reaches the desired level, it turns off the heater. Another example is a water pitcher in my refrigerator that automatically refills when placed back empty.

2. Memory-based Agents: These agents remember past experiences and use them to make decisions.

Example: A chatbot that recalls previous conversations.

How it works: The chatbot stores information from past interactions. When a user returns, the chatbot can reference earlier topics, questions, or preferences to provide personalized responses.

3. Goal-based Agents: These agents work towards specific goals, considering both the current situation and past experiences.

Example: A self-driving car navigating through traffic; think Tesla or Waymo.

How it works: The car's goal or objective is to reach a desired destination safely. It uses sensors to detect obstacles, plan routes, and make decisions to avoid collisions and traffic jams. (Note: This is one of the most fascinating types of AI agents to me.)

4. Utility-based Agents: These agents are similar to goal-based ones, but they also consider the potential outcomes of their actions to maximize "utility" or satisfaction.

Example: A financial advisor that recommends investment strategies.

How it works: The advisor factors in the client’s risk tolerance, financial goals, and market conditions to suggest the best investment options. It aims to maximize the client's return while minimizing risk. (Note: One could argue this is similar to goal-based agents.)

5. Learning Agents: These agents learn from their interactions with the environment and improve their performance over time. They can adapt to new situations and challenges.

Example: A fraud detection system in a bank.

How it works: The system analyzes transaction data to identify fraudulent activities. As it encounters new types of fraud, it learns to recognize and prevent them.

These five types represent the MOST popular categories of AI agents, though there are many others.  Now that we know the types of AI agents, the next question is,... 

"How can I use them in the sales process to increase revenue (IR), reduce cost (RC), or expand market share (EM)?"         

Next: AI Sales Agent (Part 3)

Victor Antonio | Author | Speaker

'Sales Ex Machina: How AI is Changing the World of Selling' (2018) Download a Free Copy

#aiinsales #aiagents