@aicaution Maybe some AI-skeptical boffins should invent a rigorous terminology and categorisation system so that we can describe individual systems under the "AI" marketing banner in terms of the actual technology used: The type of model, whether it is predictive or generative, language-based or numeric, provenance of training data, computation power required, open or closed. Then we will have a better framework for discussing the harms and impacts of individual products branded as "AI".