OK, I didn't agree with your article.

Symbolic AI is a backwater today. All the action is in machine learning, and that does not use a representational approach to intelligence.

Indeed, it's the lack of any representation that's a big problem for ML models, which can answer questions, but can't explain those answers, and also, cannot be "tweaked" to fix problems without retraining.

I literally cannot name one modern AI project that does use a symbolic or representational approach, and that certainly includes Tesla's inferior self-driving "AI".