Every Time I Fire a Linguist": When Statistics Replaced Understanding | The Bitter Lesson скачать в хорошем качестве

Every Time I Fire a Linguist": When Statistics Replaced Understanding | The Bitter Lesson

"Every Time I Fire a Linguist": When Statistics Replaced Understanding | The Bitter Lesson "What do I have to do to convince you that neural networks are the future?" The skeptic took a sip of his drink, buying time the way skeptics always do. "Win ImageNet," he said. The words landed with the weight of a public metric: clean, cruel, simple. Not a philosophical argument. A leaderboard. Meanwhile, in a different war, Frederick Jelinek sat in the back of an MIT lecture hall watching Noam Chomsky draw trees and arrows on a chalkboard. His wife had wanted to hear the famous linguist. Jelinek took notes—but not the way the other students did. His notes were small, surgical questions: What do we predict? What is the error? How do we know? Years later, he would say the line that cut through a generation of researchers like a blade: "Every time I fire a linguist, the performance of the speech recognizer goes up." Laughter. Shock. Silence. The numbers were improving. The systems were shipping. The world outside the lab did not care about the dignity of linguistic theory. Scoreboards have a violence that feels objective. A community will tolerate many things. It will not tolerate losing on a public metric forever. 📖 From "The Bitter Lesson" — a novel exploring 70 years of AI history through Rich Sutton's famous essay and Rumi's mystical philosophy. #ArtificialIntelligence #AIHistory #SpeechRecognition #NoamChomsky #DeepLearning #ImageNet #TheBitterLesson #MachineLearning #NeuralNetworks #NLP