TechCrunch AI Glossary: Essential Terms Explained Clearly
A living glossary by TechCrunch breaking down key AI terms including AGI, AI agents, API endpoints, chain-of-thought, compute, deep learning, diffusion, and distillation.
Stock photo for illustration only, not from the actual event
- Essential AI terms curated for developers, investors, and tech readers
- Clear definitions of AGI, AI agents, API endpoints, and chain-of-thought
- Breakdown of compute, deep learning, diffusion, and distillation
- A living document regularly updated by TechCrunch as the field evolves
As artificial intelligence continues to advance rapidly, complex technical terminology often confuses developers, investors, and everyday readers alike. TechCrunch has published a comprehensive plain-English AI glossary to help audiences easily understand the most common terms encountered across its articles and podcasts, serving as a living document updated regularly alongside the industry's evolution.
The term Artificial general intelligence (AGI) remains nebulous, yet it generally describes AI surpassing average human capabilities across most tasks. OpenAI CEO Sam Altman once described AGI as the equivalent of a median human hireable as a co-worker, while OpenAI's charter defines it as highly autonomous systems outperforming humans at economically valuable work. Meanwhile, Google DeepMind views AGI as AI matching human cognitive capabilities.
An AI agent refers to a tool utilizing AI technologies to execute multistep tasks on behalf of users, extending far beyond standard chatbots by handling expense filing, booking, or coding maintenance. Meanwhile, API endpoints act like hidden buttons on software backends, enabling external programs and AI agents to interact with services, pull data, and execute automated actions without manual user intervention.
Understanding these foundational terms is crucial because AI terminology is frequently utilized across varying corporate contexts with slightly different definitions. Having a standardized glossary bridges this knowledge gap, enabling readers and industry observers to accurately evaluate technological developments and AI capabilities.
Stock photo for illustration only, not from the actual event
Moving toward reasoning and infrastructure, Chain-of-thought reasoning in large language models involves breaking down complex problems into intermediate steps to enhance accuracy, mirroring how humans use scratchpads and equations to solve multi-variable math problems. Meanwhile, Compute shorthand represents the essential hardware infrastructure—such as GPUs, CPUs, and TPUs—that powers AI model training and deployment.
Advanced techniques also include Deep learning, a subset of machine learning featuring multi-layered artificial neural networks that autonomously extract data characteristics and learn from errors. Diffusion powers generative models by dismantling data structures with noise before mastering a reverse process to recover data. Lastly, Distillation applies a teacher-student framework to transfer knowledge from massive AI models into smaller, more efficient counterparts.
Source: TechCrunch
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment