3 ms·
"Democratizing" makes more sense when the industry evolved from AI/ML frameworks which required a Ph.D to use to allowing anyone to train a model (e.g. Theano -
by minimaxir 6y ago
"Democratizing" makes more sense when the industry evolved from AI/ML frameworks which required a Ph.D to use to allowing anyone to train a model (e.g. Theano -> TensorFlow -> Keras/fast.ai), and allowing people to train models on GPUs without a grant-funded supercomputer cluster, for very little cost (spot/preemptible cloud GPUs, Google Colab)
In this case, I agree it's not equitable.