The annual number of research papers in AI/ML started with a modest 154 papers in 2000, and the count only grew modestly through 2011. There was a sharp inflection starting next year 2012 when number of papers was almost double of the previous one. The growth noted a second resurgence beginning in 2017. This was when deep learning entered mainstream and DeepMind's AlphaGo made headlines by defeating Lee Sedol.
Note: This dataset exclusively covers research papers published on arXiv. However, this is a reasonable proxy, as the majority of computer science research papers are typically made available through arXiv.