Format Video / YouTube Embed "The predictability of programming tokens is useful not only for code synthesis but also for predicting what a developer will do next while editing existing code." Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER How Code Diverges From Prose Source title Cursor Team: Future of Programming with AI | Lex Fridman Podcast #447 Source authors Lex Fridman; Aman Sanger Source year 2024 Confidence high Importance high
Format Online paper / PDF "Language-model surprisal, already associated with cognitive processing costs in natural-language reading, also predicts aspects of human code-comprehension difficulty." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER How Code Diverges From Prose Source title Does Surprisal Predict Code Comprehension Difficulty? Source authors Casey Casalnuovo; Prem Devanbu; Emily Morgan Source year 2020 Confidence high Importance mid
Format Online paper / PDF "Functionally equivalent snippets of code with higher language-model surprisal take humans measurably longer to comprehend." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER How Code Diverges From Prose Source title Does Surprisal Predict Code Comprehension Difficulty? Source authors Casey Casalnuovo; Prem Devanbu; Emily Morgan Source year 2020 Confidence high Importance high
Format Online paper / PDF "The statistical predictability of code extends beyond programming-language syntax to patterns specific to individual software projects." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER How Code Diverges From Prose Source title On the Naturalness of Software Source authors Abram Hindle; Earl T. Barr; Mark Gabel; Zhendong Su; Prem Devanbu Source year 2012 Confidence high Importance mid
Format Online paper / PDF "Code came out more repetitive than English, with entropy sinking under two bits per token in places where English models were still hovering close to eight." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER How Code Diverges From Prose Source title On the Naturalness of Software Source authors Abram Hindle; Earl T. Barr; Mark Gabel; Zhendong Su; Prem Devanbu Source year 2012 Confidence high Importance high
Format Online paper / PDF "Back in 2012, Abram Hindle and colleagues trained statistical language models on ten Java codebases and compared their predictability with English." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER How Code Diverges From Prose Source title On the Naturalness of Software Source authors Abram Hindle; Earl T. Barr; Mark Gabel; Zhendong Su; Prem Devanbu Source year 2012 Confidence high Importance high
Format Online paper / PDF "Language models can deliver real productivity improvements on selected writing tasks without establishing the factual reliability of every generated statement." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER Can We Automate Uncertainty? Source title Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence Source authors Shakked Noy, Whitney Zhang Source year 2023 Confidence high Importance mid
Format Web Page "Scaling AI infrastructure has materially affected Microsoft’s cloud economics." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER Can We Automate Uncertainty? Source title FY25 Q4 - Performance - Investor Relations - Microsoft Source authors Microsoft Source year 2025 Confidence high Importance mid
Format Web Page "Ed Zitron links the generative-AI investment narrative to dependence on GPU sales, infrastructure spending and data-center expansion." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER Can We Automate Uncertainty? Source title The Hater's Guide To The AI Bubble Source authors Ed Zitron Source year 2025 Confidence high Importance mid
Format Web Page "A decision model can guarantee conformity to a predefined output schema without thereby guaranteeing that its chosen decision is factually correct." View Source Location PART 3 - THE THIRD AXIS CHAPTER 15: DEPLOY NOW, PRAY LATER Can We Automate Uncertainty? Source title Introducing System One Models & Jev Source authors Diogo Almeida Source year 2026 Confidence high Importance high
