I mostly agree with what you've said except for this:
but what we're calling "AI" today is basically just a spell-checker on steroids,
That's only somewhat true if you're talking about LLMs like ChatGPT.
AI itself has become a much broader term than it used to be. There are a lot of different kinds of AI out there. Generative AI like text generation (LLMs), image generation (upscaling, or creating images from scratch), or music generation (Suno). Computer Vision is another kind which can include image recognition, object detection, facial recognition, etc. And there are others beyond this.
The AI we're talking about here falls more under Computer Vision for AI which includes image recognition. In this case the machine learning model has been trained on massive amounts of images like MRIs or CT scans.