Ghost Work
The hidden human labor that powers 'automated' AI and web services, named in the 2019 book by Mary L. Gray and Siddharth Suri.
Concepts
Plain-language explanations of the ideas behind modern AI.
355 entries, all primary-sourced
The Scaling Hypothesis
The claim that broad capability emerges mainly from scaling up simple models with more compute and data.
Neuro-Symbolic AI
An approach that combines neural-network learning with symbolic reasoning to get the strengths of both.
Multi-Agent Debate
A method where several LLM copies propose and argue over answers across rounds, improving factuality and reasoning over a single pass.
Narrow AI vs General AI
The distinction between systems that excel at specific tasks and the long-sought goal of broad, flexible intelligence.
The AI reverse acqui-hire
A deal pattern where a big tech firm hires a startup's founders and licenses its technology, absorbing the team without a formal acquisition.
Agentic Workflows
Andrew Ng's framing of four design patterns - reflection, tool use, planning, and multi-agent collaboration - that make LLMs act like agents.
Many-Shot Jailbreaking
A jailbreak that defeats LLM safety training by stuffing the prompt with many fake examples of the model complying with harmful requests.
AI for science
A pattern in which learned models replace or accelerate physical simulation and brute-force search across biology, weather, materials and mathematics.
Mixture-of-Agents (MoA)
A layered method that stacks several LLMs so each refines the prior layer's outputs, beating GPT-4 Omni on AlpacaEval 2.0 with open models.
The AI bubble debate
The argument over whether AI investment has run far ahead of revenue and will deflate, or is a justified infrastructure buildout.
The AI ROI gap
The gap between massive AI investment and measurable returns - reflected in modest enterprise adoption and a revenue base far below infrastructure spend.
Foundation-model commoditization
The thesis that base AI models are becoming interchangeable commodities as open models catch up and per-token prices collapse.
Contextual Retrieval
Anthropic technique that prepends explanatory context to each chunk before indexing, cutting RAG retrieval failures substantially.
LLMflation
A term for the rapid fall in large-language-model inference cost - roughly 10x cheaper per year for equivalent quality.
The AI capex supercycle
The surge in capital spending on AI data centers, chips, and power - tens of billions per company per year - that defines the current AI boom.
Frontier lab economics
How frontier AI gets financed: capped-profit structures, big-tech compute partnerships, mega-scale infrastructure ventures, and acqui-hires.
Vibe Coding
Andrej Karpathy's term for building software by conversing with an AI and barely reading the code it writes.
Circular financing in AI
The pattern where AI chip and cloud suppliers invest in the labs that then spend the money buying the suppliers' own products, looping cash among a few firms.