What is artificial intelligence?

So this is a bigger topic than I could ever cover, but it’s still worth considering for a moment in the context of spatial AI.

What if we were to unpack “artificial” and “intelligence” independently in the dictionary:

• artifice → “clever or cunning devices or expedients, especially as used to trick or deceive others” (New Oxford American Dictionary 2023)
• intellect → “the faculty of reasoning and understanding objectively, especially with regard to abstract or academic matters” (New Oxford American Dictionary 2023)

Note that “artifice” hints at using means of “simulation” in particular. More on this later.

Another dictionary definition from the Oxford English Dictionary:

• “The capacity of computers or other machines to exhibit or simulate intelligent behaviour; the field of study concerned with this. In later use also: software used to perform tasks or produce output previously thought to require human intelligence, esp. by using machine learning to extrapolate from large collections of data.” (OED, 2024)

The first chapter of Artificial Intelligence: A Modern Approach (4th edition) provides four categorical approaches to defining AI:

1. Acting humanly: The Turing Test approach.
2. Thinking humanly: The cognitive modeling approach.
3. Thinking rationally: The “laws of thought” approach.
4. Acting rationally: The rational agent approach.

In the section on the last, “Acting rationally,” the authors Stuart J. Russell and Peter Norvig note, “In a nutshell, Al has focused on the study and construction of agents that do the right thing.”

Finally, there are some hundred definitions compiled by groups contributing to the EU’s AI Act legislation. One definition stands out to me from AI Watch Definition Artificial Intelligence, by the OECD (2019, pg 22):

• “An AI system is a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. AI systems are designed to operate with varying levels of autonomy.”

It is this last definition calls out a useful distinction, the idea that AI makes “predictions, recommendations, or decisions influencing real or virtual environments.” For me, this sets AI apart from just another form of automation, and it also calls out an influence over “environments,” key to this conception of spatial AI.

Visiting, “What is AI?”

What “artificial intelligence” actually means could easily take volumes to explore. I enjoy perusing what different authors and organizations have to say about it, since it reveals so much about what and how we think about technology and ourselves.

If we look to science fiction, we find more terrifying portrayals of AI than those in academic works or machine learning courses. But one common concept is the notion of “agents” and “environments.”

An AI System Is …

My favorite definition of an AI system so far is the following from a publication contributed during the formulation of the EU AI Act, referenced as OECD, Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449 (page 7):

An AI system is a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments. AI systems are designed to operate with varying levels of autonomy.

Note the use of “real or virtual environments.” More texts do the same.

Environments and Agents

Here, Artificial Intelligence: A Modern Approach (Russell Norvig 55), introduces the abstraction of an agent and the environment to which it relates:

The introduction to Chapter 2 says:

We will see that the concept of rationality can be applied to a wide variety of agents operating in any imaginable environment. Our plan in this book is to use this concept to develop a small set of design principles for building successful agents—systems that can reasonably be called intelligent. We begin by examining agents, environments, and the coupling between them.

And as well from Algorithms for Decision Making (Kochenderfer Wheeler Wray, Figure 1.1):

I also like the latter’s description of an “agent” in Section 1.1:

An agent is something that acts based on observations of its environment. Agents may be physical entities, like humans or robots, or they may be nonphysical entities, such as decision support systems that are implemented entirely in software.

These three texts treat “environments” as a key AI concept, although ultimately they end up mostly focusing on the intelligent agent, like a robot navigating a factory or an autonomous vehicle navigating city streets.

What role should it play?

As for the role of AI, this asks questions around the expectations of and capacity to which we should deploy it. Is it a form of automation to save us time? Is it a companion that presents us with design options and adapts to our preferences over time? Is it a creative force itself, attempting to innovate and reason?

References

• Artificial Intelligence: A Modern Approach.
• Algorithms for Decision-Making.
• OECD, Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449