Cognitive ability / Intelligence / IQ
When we’re filling a job opening with a human resource, we often submit candidates to cognitive tests, like the IQ test or the GMAT.
Why? Because we want intelligent people on our team.
But with AI Agents, it’s different.
We don’t test their intelligence.
We choose their intelligence.
As I mentioned at the beginning of this book, the “brains” of AI Agents are language models or LLMs (Large Language Models).
That’s where the artificial intelligence that brings our Agents to life resides.
And there are several of these “brains” available, each with its own capabilities and specializations.
Some are better at creative tasks, like writing and ideation.
Others are better at analytical tasks, like research and data analysis.
And some are simply more intelligent in general, capable of handling a wide range of complex tasks.
When creating an AI Agent, we select which of these brains it will have.
It’s like we could choose the IQ of our employees.
But it’s important to note that always choosing the most advanced intelligence is not always the best strategy.
Just as hiring smarter humans generally costs more, using more powerful language models also has a higher cost.
We’ll talk more about this in the topic about Cost and Salary of AI Agents.
Moreover, having more Agents with average intelligence and good resumes can, in certain cases, deliver better results than having a single super-smart Agent.
We’ll explore more of that idea when we talk about Multi-Agent Teams and Hybrid Teams of Agents and Humans.
So, how do we decide what level of intelligence our Agents need to have?
It depends on the problem we’re trying to solve, the process we’re trying to improve, or the project we’re trying to complete.
It also depends on the budget we have available and the overall strategy of our AI team.
The key here is to understand that, with AI Agents, we have direct control over their cognitive capacity.
We’re not limited by the pool of available candidates.
We can customize our Agents’ intelligence according to our specific needs.
It’s an immense power, but also a responsibility.
We need to be strategic in our choices, always considering the value that a certain level of intelligence will bring versus the cost of implementing it.
The smartest Agent is not always the answer.
Sometimes, a more balanced and diversified approach can be the best way.