Go online, and you’ll find someone talking about AI in approximately 0.2 seconds. Perhaps not since the modern smartphone has so much attention been put on one piece of technology. But unlike such things as the home computer or the smartphone, you can’t really “see” AI. It’s in machines; it’s part of software. You can’t hold it in your hands.
And maybe this is why people are often calling something AI when they don’t mean it—or they mean a different definition of the word. Despite being complicated to understand, Cryptocurrency is one thing. NFTs are one thing. The term means a specific, if nuanced, piece of technology. One thing.
By my count, the way people use the term AI are actually talking about three different things, and they are not actually—at all—similar in how they are created. So, let’s sort this out.
- Generative AI.
This is the “new” thing. This is what Claude, Grok, OpenAI, etc. are. We’ve actually had similar tech for decades, but they were nowhere close in terms of raw power. This is what a “chatbot” is. A different name for them is LLMs, meaning “large language models.” Generative AI generates things. It takes a truly massive amount of data—seriously, it’s hard to conceptualize how big a dataset these things work from—and recombines that data to generate something.
- Assistive AI
Until recently, these were the type of AI you likely interacted with the most. Assistive AI is a broad term and means a lot of different systems. But their function is usually to “assist.” Assistive AI might convert speech to text, help you with grammar, or make a task easier. They were built a completely different way than LLMs, and—though there are too many types to generalize—Assistive AIs sometimes operate on a series of pre-programed rules.
- Sci-Fi AI
This type of AI, sometimes called AGI or “artificial general intelligence,” isn’t real. Not yet anyway. If you’ve ever talked to an LLM, it may appear there’s some kind of human-like intelligence somewhere, but there’s not. It’s just knows the content of ways more conversations than you could fathom, and human communication has patterns. We currently don’t have the means to create AGIs. Some people believe that we will be able to sooner rather than later, but that’s beyond the scope of this article.
–
So, there are the three common types. If someone mentions AI at, say, a dinner party, they likely mean one of the three. And if you want to talk to them about it, it might be good to ask which one they mean. It could save a lot of time.
