Knowledge

Articles

Practical pieces on responsible AI use, written for business owners and organisations without a technical background. Further down you will also find my research and the reading list that goes with my manifesto.

NLP · July 2026

The language model you’ve been using for years without knowing it

Google BERT has understood language since 2019; ChatGPT and Claude write text of their own. On what they share, and when a small model is the better choice.

Machine learning · July 2026

Two kinds of AI, two risks: machine learning or generative AI

Machine learning predicts, generative AI creates. Knowing the difference tells you which questions to ask of an AI tool.

Architecture · July 2026

Why most AI problems are really architecture problems

When an AI project disappoints, the model gets the blame. Usually the problem is everything around it: the data, the fit with your systems, and who’s watching.

AI ethics · July 2026

Growing up in AI ethics, and why you don’t need a perfect score

Between AI principles and practice lies a gap. The maturity model from Erasmus University shows you where you stand, across six dimensions.

Humane technology · July 2026

Technology that sees people

Every technology carries its makers’ intentions in it. On humane technology, human-seeming AI, and the manifesto I hold my work to.

EU AI Act · July 2026

What the EU AI Act asks of you as a business owner (and what it doesn’t)

The AI Act sounds like something for the lawyers at big tech companies. It touches you too, though in practice it turns out to be manageable once you know what to look for.

Privacy · July 2026

Which client data do you share with ChatGPT, and which do you not?

Most data leaks through AI come not from hackers but from well-meaning prompts. Three rules of thumb you can apply today.

Research · NLP

Algorithms built to spot online hate, that turn out to be biased themselves

My master’s thesis at the University of Amsterdam: research into bias in algorithms that detect hate speech on Twitter.

Machine learning · NLP

How I labelled 1300 tweets and trained a BERT model that recognises hate speech

The technical heart of my thesis, from labelling by hand to a trained classification model, step by step in the notebook.

Reading list

The books that shaped how I see AI and humanity

The reading list that goes with my AI manifesto: books on technology, ethics and people that I recommend to anyone.