Understand how AI agents work

- So let's explore how smart AI agents work using a hypothetical example of an AI travel agent. Similar to a human travel agent, in the future, you'll be able to ask your AI travel agent to plan a vacation and it will make all of the arrangements for you. We're not there yet but we're getting close. Here's a chart that illustrates each of the steps. First, smart agents make decisions based on goals such as plan a week long family vacation to Hawaii. As you can imagine, that's a complex task. Your AI travel agent would need to know all of the steps from booking flights to finding a hotel. It also needs to take into account individual family member preferences when planning for food and activities. In this example, the agent would break down the goal into those individual tasks and figure out which ones need to be tackled first, and make a plan on what's needed to be accomplished in each task. For example, the AI travel agent may have a task list of book flights, then hotel, then plan activities. The next step the AI takes is to gather information either by asking the person making the request, tapping into a database that it has access to, or doing its own online research. For example, the AI travel agent may design a questionnaire for each family member to fill out stating their preferences. It can then do online research on options that maximizes those preferences. As the AI agent gathers information, it may learn something new and refine its approach, thus changing the prioritized list of tasks it does. For example, learning that a certain activity is available only on a certain day may require that it book flights for a different day to arrive. Once it's done, the AI agent delivers the results. For example, the AI travel agent would provide a complete itinerary for the entire trip with details like opening hours and travel time between activities. It may also come back with options or ask for clarifications to resolve a conflict between two incompatible preferences. If the requester trusts the agent, the AI travel agent may also go ahead and execute on its recommendations, book flights and hotel, buy tickets, and make restaurant reservations. It's interacting with its environment and leveraging other AI agents to get things done. Finally, the AI travel agent learns and adapts. This is the hallmark of generative AI. It can learn from each interaction and use the information to improve. For example, your AI travel agent won't know your travel preferences. In the beginning, it may be asking you many questions, seeking to understand you better. But over time, it will know that you prefer aisle seats, which airlines to use, and even suggest new activities you may like to try. Here's a final thought on the importance of trust. The power of AI agents rests in our ability to trust that they will accomplish the objectives given to them without our intervention. That trust develops through multiple interactions over time. So now you have a high level understanding of how a smart AI agent can work. In the next video, we'll explore the different types of AI agents, how they create business value, and their limitations.