The headlines are hard to miss: AI could kill all humans in 10 years, a growing number of AI researchers and executives have warned in recent weeks. Not everyone believes it’s as dire as some are making it out to be, and instead see much of the hype as a marketing campaign designed to fill the pockets of Big AI. Regardless of where you are on the contentious issue, one thing is clear: AI poses enormous risks, and whether regulation or a slowdown happens or not, the technology itself isn’t going anywhere and will likely continue to change the way we live.
That’s why it’s important to understand the confusing word salad of phrases and acronyms researchers and experts often use to describe the fast-developing technology, its risks, and its potential benefits.
To help, we’ve put together a quick cheat sheet of nine of the most common confusing terms – from AGI to ASI to RSI to “alignment” – you should know. Many of these terms don’t have universally agreed-upon definitions and refer to theoretical concepts that some experts don’t even think are possible, but here’s how best to think about them:
1. Artificial General Intelligence (AGI)
While popular AI applications like Claude or ChatGPT can be impressively capable with specific sets of tasks, artificial general intelligence (AGI) would match or exceed human capabilities across different areas. There is no single definition of AGI, but you can think of it as an AI system that can learn, solve problems, and generally operate much like – or better than – a human. People can have relationships with others and make their own decisions. Chatbots, while impressive, are only able to generate text that’s similar to what they’ve been trained on, based on a prompt. A theoretical AGI model would have the same flexibility that humans have.
AGI remains a hypothetical term. But OpenAI CEO Sam Altman says his company will reach AGI by the end of the year. Anthropic CEO Dario Amodei, however, says AGI is just a “marketing term,” though he has admitted that we are on the verge of creating AI systems that are “better” than humans at most things.