In this video, TechTarget editor Jamison Cush explains the four main types of AI.
Do You Really Know? artificial intelligence Is that so? Well, that may depend on the context.
AI is a spectrum, and experts consider different parts of it. narrow or weak And something closer to true general artificial intelligence, or strong AI. Here we discuss four main categories of AI:
AI can be divided into four categories: reactive, limited memory, theory of mind, and self-aware. Let's discuss each of them.
The first is reactive AI. Most machine learning models Reactive AIThese models use statistical mathematics to account for vast amounts of data and produce seemingly intelligent outputs.
Reactive AI is good for simple classification and pattern recognition tasks – for example, it was thanks to reactive AI that chess grandmaster Garry Kasparov was defeated in 1997. Chess is a game of numbers, and AI models can calculate all possible moves faster and more accurately than humans can.
Or think about Netflix: While it may seem like Netflix has an idea of what content you want to watch, it's all just calculations, using the purchasing history of other customers to create their recommendation algorithms.
But reactive AI cannot handle the imperfect information that humans, by comparison, can handle with ease. Humans are masters of prediction, and they deal with imperfect information all the time.
The second type are machines with limited memory. Currently, AI developments such as conversational AI chatbots and self-driving cars fall into this category. Machines with limited memory use deep learning, which mimics how humans acquire knowledge. This type of AI can handle complex classification and make predictions using historical data.
Although machines with limited memory can far outperform humans at certain tasks and can autonomously improve at these tasks over time, they are still considered limited AI because they require vast amounts of training data to learn tasks that humans can learn with just a few examples. For example, if you give a toddler a few pictures of a cat, it can roughly understand what a cat is. But a machine with limited memory would need thousands of pictures to learn the same thing.
The third is Theory of Mind AI, which is hypothetically able to understand human motivations and reasoning. Also known as artificial general intelligence, it can learn from fewer examples than machines with limited memory, contextualize information, and apply that knowledge to a wide range of problems.
Theory of mind is still a long way away due to AI's main limitation: understanding. Currently, AI models can accurately write essays and books, but they cannot understand what they create. A theory of mind AI can.
4. Self-Aware AI – While theory of mind AI can understand the emotions of others, self-aware AI refers to a system that is aware of not only its own internal state but also the internal states of others, i.e. a machine that can match human intelligence and emotions.
This is sometimes called artificial superintelligence, but the irony is that we don’t even understand this level of AI because we don’t understand the human brain well enough to build an artificial one.
Do you think general artificial intelligence is possible? Should we pursue it? Share your thoughts in the comments and don't forget to like and subscribe.
Sabrina Polin is the Editor-in-Chief of Video Content for the Learning Content team. She curates and develops video content for TechTarget's editorial YouTube channel, Eye on Tech. Previously, she was a reporter on the Product Content team.
