THE AI NOTEBOOK / AI EXPLAINED
AI Explained
Clear explanations of the technology behind the answers. Start with the question you need answered. Where a subject takes more than one article, each part has its own focus.
What to expect
A short summary first, followed by an accessible explanation and a practical conclusion. Geek Boxes provide extra detail where it helps. Some articles will include a “Try this” exercise, with an explanation of what the result can and cannot show.
Browse by topic
Articles appear under each topic as they are published.
How AI Works
Learning, memory, search, reasoning and why an AI’s tone changes.
-
How AI Learns
Download this article (PDF) AI Explained By Dr Peter Fields THE 2-MINUTE SUMMARY A language model works with tokens, small pieces of text represented as numbers. During pre-training it learns…
-
Knowledge, Search and Tools (How AI Knows Things)
Download this article (PDF) AI Explained By Dr Peter Fields THE 2-MINUTE SUMMARY An AI answer can draw on patterns learnt during training, information you supply, or data returned by…
-
Memory, Attention and Limits (Why AI ‘Forgets’)
Download this article (PDF) AI Explained By Dr Peter Fields THE 2-MINUTE SUMMARY The conversation you can scroll through and the information supplied to answer your next message are two…
-
Reasoning, Speed and Effort (How AI ‘Thinks’)
Download this article (PDF) AI Explained By Dr Peter Fields THE 2-MINUTE SUMMARY Some AI models generate intermediate steps before producing a visible answer. A system can also spend extra…
-
Tone, Personality and Alignment (Why AI Acts the Way It Does)
Download this article (PDF) AI Explained By Dr Peter Fields THE 2-MINUTE SUMMARY The assistant’s manner is shaped by what it learnt during training, how it was adapted to follow…
Using AI Well
Prompts, confidentiality, AI agents and choosing between open and closed models.
-
Can I Put a Client’s File into AI?
Sending client material to an AI service without authority can breach confidentiality even if nobody opens the file. Find out who could read it, for what purpose and under what…
-
How to Talk to AI
The AI replies to the information the app puts in front of it. That can include earlier messages, saved preferences and material from files or search.
-
Open Weight v Closed Weight Models, Part 1 – What an open model gives you
Weights are the numbers a model learns during training. An open-weight release makes those numbers available to download under a licence.
-
Open Weight v Closed Weight Models, Part 2 – Running AI in-house and using your own documents
In-house hosting can mean the firm’s own servers or a cloud account it controls. The model is one part of the setup.
-
What Are You Asking the Agent to Do?
An AI agent can use tools and decide what to do next in light of what it finds. How much discretion it has depends on the task and the software…
-
What Did You Give It Permission to Do?
Part 1 asked what you would delegate. Here the question is what the agent can actually do.
-
Who Can See What You Share?
The app may keep your questions, files and replies so you can return to them. What it saves and what the model provider keeps are separate questions.1,2
Checking AI’s Answers
Hallucinations, sources, benchmarks and the limits of recognising AI-generated work.
Hallucinations
Read the six parts in order, from why AI makes things up to how to use it without being misled.
- Why AI Makes Things Up
- Why Doesn’t It Just Say “I Don’t Know”?
- What Has Changed Since ChatGPT First Appeared?
- It Has Sources. Can You Trust It Now?
- What Does a Hallucination Rate Actually Tell You?
- Can You Use AI Without Being Misled?
More on checking AI’s answers
-
Can You Tell It’s AI?
A text detector looks for patterns in a finished piece. Its score depends on the tool and the material it was tested on.
-
How to Read AI Benchmarks
A benchmark measures performance on a defined task under stated conditions. An exam score, a count of software fixes and a league table based on user votes tell you different…
Images, Voice and Video
How AI generates images, speech and video, and what to do about deepfakes.
-
How Deepfakes Are Made and Why We Believe Them
AI can imitate a voice, change what a person appears to say on video, or generate a scene that never happened. A genuine clip with a misleading caption can cause…
-
What to Do When Someone Uses Your Face or Voice
Call 999 if you’re in immediate danger. If someone is using your voice to ask for money, warn the people they may contact.
-
When the Sheep Starts Moving
A generated voice has to get the words, the sound of the speaker and the delivery right. A generated video has to keep a scene consistent as things move.
-
Why Did AI Move the Window?
An image generator has to make the whole scene hang together. Some models do that by working from noise towards a picture.
Arguments, evidence and consequences
For investigations and connected arguments, browse the commentary or choose an article series. For courts, legal work and access to justice, see Legal AI.