AI is rewriting the world and inventing a whole new vocabulary while it does it. TechCrunch AI just published a living glossary of the terms you’re most likely to hear in a pitch, a product meeting, or a podcast, and it’s built for people who want plain-English definitions without feeling behind. Here’s a walkthrough of that glossary, in order, with a bit of extra context on why each term matters.
Quick Start
What you’ll learn: eight core AI terms that show up constantly right now, and what each one actually means. What you need: nothing but curiosity. No coding, no math, no prior AI knowledge. Read it top to bottom or jump to the term that tripped you up.
The terms, in order
- Artificial general intelligence (AGI): A nebulous term, as TechCrunch AI puts it, that generally refers to AI more capable than the average human at many or most tasks. The definitions vary by who you ask. OpenAI’s Sam Altman once called it the “equivalent of a median human that you could hire as a co-worker.” OpenAI’s charter frames it as “highly autonomous systems that outperform humans at most economically valuable work.” Google DeepMind says it’s “AI that’s at least as capable as humans at most cognitive tasks.” Why it matters: even the experts don’t agree, so when someone claims AGI is near, ask which definition they’re using.
- AI agent: A tool that uses AI to perform a series of tasks for you, going beyond a basic chatbot. Think filing expenses, booking a table, or writing and maintaining code. The term means different things to different people, and the infrastructure to fully deliver on it is still being built. The core idea: an autonomous system that may draw on multiple AI models to carry out multistep tasks.
- API endpoints: Think of these as hidden “buttons” on the back of a piece of software that other programs can press to make it do things. Developers use them to build integrations, like letting one app pull data from another. Why it matters: as AI agents get smarter, they can find and press these buttons themselves, which opens up powerful and sometimes unexpected automation.
- Chain-of-thought reasoning: Some questions you answer instantly. Others need a pen and paper. If a farmer’s chickens and cows have 40 heads and 120 legs, you write an equation to get the answer (20 chickens, 20 cows). For large language models, chain-of-thought means breaking a problem into smaller intermediate steps. It takes longer, but the answer is more likely to be right, especially for logic or coding. Reasoning models are built from traditional LLMs and tuned for this through reinforcement learning.
- Coding agent: A more specific idea than a general AI agent, applied to software development. Instead of just suggesting code for a human to paste in, a coding agent can write, test, and debug on its own, working across entire codebases and pushing fixes with minimal oversight. TechCrunch AI compares it to hiring a very fast intern who never sleeps. The catch: like any intern, its work still needs a human review.
- Compute: Shorthand for the computational power that lets AI models run. It fuels the whole industry, both training and deploying models, and it usually points to hardware like GPUs, CPUs, and TPUs. Why it matters: compute is the bottleneck and the cost center behind nearly every AI headline about spending and chips.
- Deep learning: A subset of machine learning where algorithms use a multi-layered, artificial neural network structure loosely inspired by the human brain. That structure lets them spot important features in data on their own and learn from their own errors. The trade-off: deep learning needs huge amounts of data (millions of points or more) and takes longer to train, so it costs more.
- Diffusion: The tech behind many art, music, and text generators. Inspired by physics, diffusion systems slowly add noise to “destroy” the structure of data like photos or songs, then learn to reverse that process to create something new.
What to do next
Bookmark the original glossary. TechCrunch AI updates it regularly, so treat it as a reference you revisit, not a one-time read. Next time a term stumps you in a meeting or article, look it up instead of nodding along. And when someone drops a buzzword like AGI, ask which definition they mean. You’ll find the full, evolving list at the original source.