My stance on AI

Last updated: 07 November 2026

Machine learning is a broad area that's been around for decades, and now covers everything from social content algorithms, to image recognition, medical analysis, as well as the chatbots and generative text and image generators that everyone now unfortunately lumps together and broadly includes under the label of "AI". I have no issue with machine learning as a concept, nor do I take issue with ethical, open-source tooling produced by governments and publicly funded researchers, particularly as machine learning and supercomputing has improved and supports the lives of all beyond what one could write here. Rather notably in the areas of agriculture, environmental prediction and climate forecasting, accessibility and healthcare. However, as a traditionally trained and degree educated 3D artist, Animator and Game developer whose industry has been slowly hollowed out by "Generative AI" machine learning tools, I am not interested in supporting the use of or actively providing "Generative AI" tools for creative and artistic works. So long as I remain the developer of Thingway, it cannot, and will never write anything for you.

What Thingway will not, and can not do

There is no "AI" chat model, assistant or official integration within Thingway, and there never will be. Nothing in the app inventively writes, autocompletes, advises on or summarises any written text. There are no chat functions, and nothing to prompt.

The app is designed exclusively to make the human experience of writing digitally feel better, not easier.

Where machine learning is used

The Thingway app uses machine learning in one area, and within a solely assistive capacity, exclusively regarding machine text translation.

Translation

Thingway Premium optionally includes a downloadable, offline and on-device machine translation model that covers 35 European languages. The system is built around SalamandraTA, an open-source machine translation model developed by publicly funded researchers at the Barcelona Supercomputing Center in Spain.

This was included specifically upon the request and advice of local teaching staff who described often having students in their cohort who simply can't read the English teaching resources made for them.

Why It's not "AI" as we know it, and why you shouldn't consider it that either.

Machine learning as previously described, has by most people been effectively completely merged with the corporate marketing label of “AI", which itself has become synonymous with American Megacorporations, environmental devastation, vast generative datacentres and the theft of creative works. The provided machine translation tool simply isn't that. While it uses Large language model "LLM" architecture, in Thingway it is not a broad-spectrum "generative AI" that can be user prompted, it cannot invent translated text on its own without a hand-written source to transform and translate from, nor is the model data trained through unlawful textual works. It's a continuation of machine-translation research and tooling that's been ongoing in the field since the 1950s, in this case its publicly funded, and developed at the Barcelona Supercomputing Center (BSC), which has existed since 2005, using textual data used lawfully under European law. It runs directly on your own device without an internet connection of any kind, so once downloaded, the only energy it uses is the battery you’ve already charged. Thus I describe it as exactly what it is: a small, open-source and on-device machine translation model.

While I have no affiliation with the center, I would encourage those interested or concerned to take a further look directly at the center and translation model for themselves

Update log

07 November 2026