Glossary
M
26 entries beginning with M, the same definitions the articles use.
Machine learning
A branch of AI in which systems learn patterns from data rather than following explicitly programmed rules; the foundation of most enterprise AI capability.
Related: The Board in the machine
See also:Cloud computing
Maturity Arbitrage Concept
The strategic advantage of combining decades-proven AI disciplines like machine learning and computer vision with newer generative capabilities to balance risk while still capturing innovation at the edges.
First introduced in: The Return of Traditional AI: Organisations Are Rethinking Their LLM-First Strategies
Related: The Compound Loop: Why Agentic AI's Real Power Lies Beyond Generative AI
See also:Computer visionMachine learning
Maximum Fidelity Asset
The evidence discipline of grading every reading a Board relies on by one of four indicator types: lagging indicators of past outcomes, leading indicators of early signals, predictive indicators of future value, and reasoned indicators that prove what must hold true. It matters because a decision is only as strong as the fidelity of the evidence beneath it, and the type tells a director exactly what quality of evidence they are looking at.
See also:Automated reasoningFormal verificationLagging indicatorLeading indicatorPredictive indicatorReasoned Indicator
MCP DNS Registry Architecture Concept
A proposed standard for agentic AI infrastructure: a DNS-grounded registry through which AI agents discover, verify, and trust MCP servers across organisational boundaries, so agent-to-system connectivity scales on public-internet patterns rather than bespoke, per-vendor integration.
First introduced in: MCP DNS Registry Architecture
MCP server
A defined intermediary sitting between the agent and an underlying system, handling translation and creating a clear, auditable control point for what the agent can read or do.
Related: MCP Explained: The Agent Infrastructure Standard Boards Need to Understand
meta data
Invisible page content, typically a title, description, and keywords, sitting behind a web page to help search engines locate and rank it, linking the information creator to the information user.
Metered paywall
A subscription model, pioneered by the Financial Times in 2007, allowing readers free access to a set number of articles before requiring payment, balancing openness with revenue sustainability for digital journalism.
Related: From Print to Web to AI: Creating Sustainable Value in the AI Era
Migration bubble
The required initial investment to complete a migration, including dual-running of old and new environments, training, additional licensing and data costs, experienced regardless of whether the destination is cloud or on-premises.
Related: Planning a cloud migration? Here's what you should consider
See also:Training
Minimum Lovable Governance Asset
Governance embedded in how work happens: proportionate to risk, continuous rather than episodic, and used because it works. The operating principle through which duties like the DUAA safeguards are actually delivered.
Related: AI Centre of Excellence: The Essential Functions of the Five PillarsShadow AI amnesty: from discovery to governance
See also:Shadow AISix Board ConcernsData (Use and Access) Act 2025AI Amnesty
Minute-Fidelity Failure Concept
When AI-drafted minutes capture what was said but not what was contested, weighed or dissented from, leaving an official legal record no one can fully defend.
First introduced in: AI and the Company Secretary: Operating the Boundary the Chair Polices
See also:Summary Substitution Failure
Mixture-of-experts
An efficiency technique that activates only relevant sub-models per query, one of several approaches (alongside distillation, quantisation, and speculative decoding) delivering compute reductions along the inference pipeline.
Related: The Headroom Argument: Why AI Efficiency Means More Compute, Not Less
See also:Inference
MLOps
Machine Learning Operations; standardised operational practices for consistent model serving, monitoring for drift and performance degradation, enabling reliable AI operation at scale rather than handling each initiative as a separate project.
Related: AI Centre of Excellence: The Essential Functions of the Five Pillars
See also:Machine learning
model card
A living document produced before deployment recording what an AI system is intended for, what it has been validated against, and where it is known to be unreliable or limited. It is the artefact a director asks for when the question is not what a system can do but what it has been shown to do safely.
Related: Increasing AI Maturity: Navigating the AI Stages of Adoption with the Five Pillars
Model Context Protocol
An open infrastructure standard, likened to USB, providing a single standardised interface that connects AI agents to proprietary enterprise data, systems, and processes without bespoke integrations.
Related: MCP Explained: The Agent Infrastructure Standard Boards Need to Understand
Model Documentation Form
A standardised disclosure covering an AI model’s capabilities, limitations, training-data provenance, compute usage, and energy metrics, with a ten-year retention requirement under the GPAI Code.
Related: Why Boards Need to Watch the EU's General-Purpose AI Code of Practice
See also:Training
Model drift
The gradual degradation of an AI model’s performance over time as real-world conditions diverge from its training data, cited as a novel AI risk category requiring specialist expertise to detect.
Related: Increasing AI Maturity: Navigating the AI Stages of Adoption with the Five Pillars
See also:Training
Model extraction attacks
A security threat in which competitors reconstruct a proprietary AI model; cited as a reason security controls must go beyond access management to protect valuable embedded research insights.
Related: AI Centre of Excellence: The Essential Functions of the Five Pillars
Model provenance
Understanding how a model was trained, on what data, and under what licence, a fundamental risk consideration affecting intellectual property exposure and long-term costs.
Related: Selecting your enterprise LLM: Moving beyond the hype to make the right choice
Money for Old Rope Concept
Transforming forgotten or seemingly worthless legacy assets into new revenue streams, a phrase coined during a 1998 newspaper archive digitisation project and applied since to data and AI opportunity.
First introduced in: Introducing the AI Stages of Adoption: A framework for understanding AI readiness in your business
See also:Data monetisation
Multi-agent systems
Swarms of specialised AI agents that collaborate, compete and produce emergent behaviours, shifting governance from controlling individual models to orchestrating entire agent ecosystems and their interactions.
Related: AI Centre of Excellence: Future-proofing Through Continuous Evolution
See also:AI agent
Multi-currency transactions
Ecommerce functionality enabling customers to pay in their local currency, requiring attention to banking settlement, exchange rates, local taxation, pricing and financial reporting rather than simply displaying prices in different currencies.
Multi-Speed Adoption Concept
The reality that different business functions adopt AI at markedly different rates and maturity levels along the AI Stages of Adoption, requiring tailored governance rather than uniform, one-size-fits-all policies.
First introduced in: Orchestrating Multi-Speed AI: The Complete AI Framework as Guiding Policy
Related: Rethinking Business Cases in the Age of AI: Creating the Foundation
Multi-Speed Collision Concept
When functions align AI to their own objectives at different velocities, their individual successes actively undermine each other, for example marketing generating demand that supply chain AI cannot fulfil.
First introduced in: AI's Interconnected Challenge: Diagnosing the Six Concerns of the Board
See also:Multi-Speed Governance
Multi-Speed Governance Concept
Governance recognising that different business functions adopt AI at different speeds and maturities simultaneously, applying varying oversight intensity rather than uniform, one-size-fits-all control that either stifles or fails to contain risk.
First introduced in: AI Centre of Excellence: Designing Structure for Multi-Speed Governance
Multi-Speed Reality Concept
Different parts of a business sitting at different AISA stages simultaneously, marketing transforming with AI content while operations experiments with predictive maintenance, demanding coordination rather than a single organisation-wide posture.
First introduced in: AI Centre of Excellence: Mapping Your Multi-Speed AI Reality
Related: Understanding the AI Stages of Adoption: A framework for business leaders
See also:AI Stages of AdoptionMulti-Speed GovernanceMulti-Speed Adoption
Multimodal AI
AI that works across multiple data types at once, for example text, images, and audio, within a single model.
See also:Generative AIFoundation model
No M entries match. or try the site search.