What is Artificial Intelligence

Artificial Intelligence
This is a word cloud generated from the most common words on https://en.wikipedia.org/wiki/Artificial_intelligence. It was made by WordCloud Python library.

By Justin Kempf

What Exactly is Artificial Intelligence

Throughout this podcast I have talked to writers about technology about its effects on democracy. Recently the background readings have started to touch on artificial intelligence more and more. For me it raises a lot of questions, because artificial intelligence is never as intelligent as advertised. Even the most sophisticated forms of AI have limited versatility. Almost all forms of AI so far also make a fair number of mistakes, although those issues will likely decrease as the technology improves. But it still raises an important question. What is it?

I have come across artificial intelligence that writes papers, makes art, and replicates voices, but none of this explains why it involves artificial intelligence. The applications also cover a wide range of possibilities so it does not make intuitive sense why they all fall into the same basket of technologies. Some people might think this is a silly thought. They likely think of artificial intelligence as any form of cutting edge technological breakthrough. And in some ways they are right. Most of the cutting edge technological break throughs involve artificial intelligence, but that does not explain what makes it different from earlier forms of computer programming.

What is Machine Learning

The most common attribute people give to artificial intelligence is machine learning, but I fell this is disingenuous. Machines do not learn the same way humans do. Too often we have given human attributes to artificial intelligence to make it sound more human than it really is. Machine learning is the most pronounced example of this. I prefer to substitute pattern recognition for machine learning. In other words, machines learn through recognizing patterns. They learn to speak through recognizing patterns in speech and learn to write through patterns in other people’s writings. They recognize images and objects  through patterns in other images.

In some ways, this is how humans learn as well. However, it’s a very narrow form of discovery. Moreover, it’s a highly mathematical way to learn. This makes it programmable. In other words, artificial intelligence is not magical. It ultimately comes down to mathematical concepts and ideas. But what makes artificial intelligence so scary is how much humans do can get reduced down to a series of patterns. Perhaps the most frightening involve more creative forms of expression like visual art, writing, and music. Artists have complained how artificial intelligence has appropriated their art in ways that violate copyright protections. Machines discovered the patterns in the work of genuine artists and simply learned to replicate it.

Pattern Replication

What impresses so many about artificial intelligence is not simply that it discovers patterns, but can replicate them. It discovers the patterns in our voices and replicates them in voice generator programs. It discovers the patterns in writings and replicates them through chat bots. The more complex forms of artificial intelligence discover multiple forms of patterns to replicate more complicated tasks. However, no matter how complex the artificial intelligence becomes it relies on patterns for its applications. Even something as complex as General AI merely ties together multiple forms of pattern recognition and replication to perform multiple applications and functions.

Still, it’s important to distinguish artificial intelligence from more basic algorithms. A lot of work involves complex decision trees. People look at problems and work through possibilities in their head more or less intuitively. Some people recognize those patterns and take a more systematic approach. A basic algorithm can take the systematic approach and simply work through a decision tree. However, it might miss the nuances and exceptions that a skilled human would recognize. Artificial intelligence would look for the patterns in those exceptions to build an increasingly complex and nuanced algorithm to guide its approach.

Is it Scary

Ultimately, artificial intelligence is both amazing and frightening. Part of the fear comes because we do not yet understand its limits. But as we understand the technology better, we will likely have a better sense of its drawbacks and failures. In many ways, artificial intelligence offers a test of how much of the world we can understand through mathematical concepts. Those who believe we can distill the entirety of the universe into numbers will believe artificial intelligence will make human intelligence irrelevant. However, those who believe the universe is more than a series of equations will believe AI has defined limitations.

Social scientists have found ways to analyze human behavior using statistical relationships, but it’s simply not possible to predict large scale sociological behavior with perfect certainty. There is always room for doubt. Moreover, this is not simply due to a lack of information. Uncertainty is built into the formulas and equations they use to analyze behavior. Even physics finds it is impossible to make precise predictions at the quantum level. It’s not considered a failure of measurement or lack of information. They have actually codified it into a scientific principle.

Ultimately, artificial intelligence will always require human oversight to give it guidance. Human intelligence does more than recognize patterns. Humans impart values and meaning into ideas. We tackle uncertainty and suboptimal problems through difficult tradeoffs that force us to relitigate our values and priorities in ways that change our circumstances and ourselves. Artificial intelligence can become a tool to remove some of the uncertainty behind our decisions and allow us to solve unimaginable problems, but it will not eliminate the need for humans to make difficult decisions. This is why AI will not bring an end to democracy.

About the Author

Justin Kempf manages this blog and hosts the podcast Democracy Paradox. He lives with his family in Carmel, Indiana.

Call for Writers

Do you want to publish a post on the blog? Send your submissions to jkempf@democracyparadox.com.  The blog is open to publishing a wide variety of perspectives on democracy, democratization, and world affairs. But please keep submissions between 500-1,000 words.

Democracy Paradox Podcast

Make a one-time Donation to Democracy Paradox.

Become a Patron!

Srdja Popovic on Dilemma Actions

Wendy Hunter on Lula, Bolsonaro, January 8th and Democracy in Brazil

More Episodes from the Podcast

Democracy Paradox is part of the Amazon Affiliates Program and earns commissions on items purchased from links to the Amazon website. All links are to recommended books discussed in the podcast or referenced in the blog.

Leave a Reply

Up ↑

%d bloggers like this: