162 articles
Browse guides
A model card records intended use, out-of-scope use, training data at a high level, evaluation results, and known limitations. Its most valuable section is the one teams skip: the limitations. It has moved from a research norm toward a compliance expectation under both NIST and EU frameworks. In practice: ‘Evaluated on English only. Not validated for medical decisions.’
MCP standardises how models discover and call external capabilities, so a tool built once works across any compatible client instead of being rewritten per vendor. Introduced by Anthropic in late 2024 and adopted more broadly since, it is the closest thing the agent ecosystem has to a common connector. In practice: One MCP server for your ticketing system, usable from any MCP-aware assistant.
Drift happens because reality changes: customer behaviour shifts, vocabulary shifts, a competitor launches. The model is unchanged, but its accuracy quietly decays. The danger is that it fails without erroring, so only monitoring catches it. In practice: A demand forecast trained pre-inflation that keeps returning confident, wrong numbers.
Multi-agent setups split a job across agents with narrow roles — research, draft, critique — on the theory that focused context beats one agent juggling everything. They genuinely help on tasks that decompose cleanly. They also multiply cost, latency, and failure modes, so reach for one only after a single agent has demonstrably failed. In practice: One agent gathers sources, a second writes, a third fact-checks against the sources.
Multimodal models map different data types into a shared representation, so you can hand one a screenshot and a question and get a text answer. The practical upshot is that whiteboards, PDFs, and voice all become valid inputs, which removes the transcription step from a lot of workflows. In practice: Photographing a broken chart in a slide deck and asking what is wrong with it.
Narrow AI performs well inside its intended scope and does not transfer outside it. The label is a contrast with AGI rather than a slight — narrow systems are the ones creating real value today. A general-purpose model can look broad and still be narrow in the sense that matters: it does not act, learn, or adapt beyond its design. In practice: A model that writes excellent Python cannot drive your car, however fluent it sounds. ...
Negative prompts give a diffusion model an explicit steer away from certain features — extra fingers, watermarks, a particular style. They are a first-class control in image tools and mostly not a thing in text models, where saying what you do want beats listing what you do not. In practice: Negative prompt: ’text, watermark, blurry, extra limbs'.
A neural network takes an input as numbers, multiplies them by learned weights, applies a non-linear function, and passes the result to the next layer. Repeat this enough times with enough data and the network can approximate remarkably complex relationships. The ’neurons’ analogy is a historical label, not a claim about biology. In practice: An image classifier is a neural network that turns 200,000 pixel values into one number per possible label. ...
Published by NIST in January 2023, the AI RMF is guidance rather than law, and it is sector-agnostic. Its four functions give teams a structure for identifying context, measuring risk, and acting on it. It has become the de facto reference for AI governance programmes, including at companies with no US obligations, because it is practical and free. In practice: Using Govern-Map-Measure-Manage as the skeleton of an internal AI policy. ...
On-device inference keeps data local, works offline, and has no per-token cost — genuinely attractive for privacy and regulated contexts. The trade is capability: local models are smaller and slower than frontier APIs. Quantization and distillation are what make it viable at all. In practice: Transcribing a confidential meeting without the audio leaving the laptop.