With the rampant rise of irrational #AI enthusiasm over the past decade, just about every candidate whom I had interviewed on my client’s behalf has claimed to be a data scientist specialising in AI.

All of them are proficient users of PyTorch, Karas, and other Python API. But so far, not one of them has ever read the original inventors’ peer-reviewed papers on the various models and algorithms of which these kids claim to be the experts. And of course none of them has ever implemented even a simple algorithm, like backpropagation, straight from Rumelhart’s seminal paper.

I am not suggesting that everyone must implement a 500 billion parameter LLM. But I am expecting that those who claimed to be an AI expert ought to have read at least the seminal papers on the topics over which they claim dominance.