TechCrunch publishes comprehensive AI glossary explaining key terms

TechCrunch has published a detailed glossary of essential AI terminology, aiming to demystify jargon like LLMs, RAG, RLHF, and AGI for a general audience. The glossary defines AGI as AI surpassing human capability, with OpenAI's Sam Altman calling it a "median human co-worker" and Google DeepMind viewing it as AI at least as capable as humans in most cognitive tasks. It explains AI agents as autonomous tools performing multistep tasks, and coding agents as specialized versions for software development. The glossary covers chain-of-thought reasoning, diffusion models, distillation, fine-tuning, GANs, hallucinations, inference, large language models (LLMs), memory caching, neural networks, open source vs. closed source, parallelization, RAMageddon (RAM shortage), reinforcement learning, tokens, token throughput, training, transfer learning, weights, and validation loss. It notes that RAMageddon is causing price surges in gaming, consumer electronics, and enterprise computing. The glossary is regularly updated.