Latest posts from @karpathy on X (Twitter)
What they post about
Andrej Karpathy's account focuses on the application of large language models (LLMs) in argumentation and software development. He discusses the evolution of integrated development environments (IDEs) and the challenges faced by programmers in adapting to new tools and abstractions. Additionally, he shares personal experiments related to fitness and health tracking. His insights reflect a deep engagement with AI technologies and their implications for programming and personal productivity.
Recent posts
-
tweet
HackerTube Offers an Alternative to YouTube Recommendations
Frustrated with YouTube recommendations, the creator built HackerTube using Claude to analyze 43 million Hacker News comments and identify 6,454 repeatedly recommended videos. The project offers recommendations without a login, uses frames from videos as thumbnails and makes its underlying data available for others to explore.
View on X (Twitter) → -
tweet
Rooted Samsung TV Raises Always-On Microphone Concerns
A repost claims that a rooted Samsung smart TV revealed an always-on microphone, raising questions about privacy and the economics of low-cost televisions. The excerpt does not provide further technical details.
View on X (Twitter) → -
tweet
Post Warns of Widening Gap in AI Understanding
A commentator argues that public understanding of large language models is becoming increasingly uneven. Their rough estimates suggest most people have little exposure to the technology, while a much smaller group uses advanced models professionally for mathematics and coding.
View on X (Twitter) → -
tweet
Testing AI’s Sense of Geography
The post describes an evaluation that asks a language model whether 16,200 latitude-longitude coordinates fall on land or water, then plots the answers as an image. The author suggests the results show geographic knowledge learned from internet data.
View on X (Twitter) → -
tweet
Making Language Model Outputs Easier to Understand
A post suggests several ways to make language model responses easier to interpret: request clearer writing using the ASD-STE100 standard, ask for diagrams, or have the model create interactive HTML pages. It also points to custom explainer videos as a promising format.
View on X (Twitter) →
30 more posts indexed for @karpathy — they go out in the digest.
@karpathy posted 1 time last week.
This page shows what they already posted. Get tomorrow's — and the day after — summarized in one morning email. First 7 digests free, no card required.
Follow @karpathy by email →
