Artificial Intelligence models may have a few issues, algorithms don't

To be fair, new technologies must be questioned, probed, and "problematized" (to use one of their favourite buzzwords) -- and it is indeed a democratic prerogative. That said, there seems to be persistent confusion around the very basics of the discipline. No other example demonstrates this best than the conflation of objectives, algorithms and models. Simplifying a little, the life cycle in creating a machine learning model from scratch is the following. The first step is to set a high-level practical objective: What the model is supposed to do, such as recognising images or speech.