coded language

Rob Miles explains on Computerphile how language models generate plausible text and how transformers improved on recurrent networks, with a look at what comes next.

A short cartoon explaining how language models like ChatGPT work, from autocomplete and conditional probabilities to neural networks, gradient descent and backpropagation. It shows how networks act as universal function approximators.

Tom Scott, with linguists Gretchen McCulloch and Molly Ruhl, examines whether the language you speak changes how you think. The short answer is mostly no, with caveats about the weak version of the Sapir-Whorf hypothesis.

3Blue1Brown's visual explanation of how transformers and GPT-style language models work, tracing data through tokens, embeddings and attention.

Vox explains the Berlin and Kay research showing that languages around the world tend to invent color words in the same order. It walks through the hierarchy and the debate around it.

Tom Scott makes a short descriptivist argument that language keeps changing and that casual internet writing conventions, like lowercase typing and dropped full stops, are legitimate. It is part of his Language Files series, written with Gretchen McCulloch.

Yuta asks whether native Japanese speakers can actually write kanji by hand, showing that many struggle to recall characters from memory.

The Map Men explain why British place names are so hard to pronounce, tracing it to layers of Viking, Anglo-Saxon, Celtic and Norman influence, with a pronunciation guide.

Computerphile video in which Mike Pound explains a University of Maryland paper on watermarking LLM output. The method biases token choices so generated text can later be detected statistically.

An animated introduction to geometric algebra, presented as a language that can describe most of physics. It was made with manim and covers multivectors and how they generalize familiar tools like cross products and quaternions.

Linguist Simon Roper reconstructs how a London accent would have sounded from the 14th to the 21st centuries, reading passages in period pronunciation. It draws on the Cambridge History of the English Language.

Rick Beato argues that adults cannot develop true perfect pitch, which he ties to critical-period brain plasticity and language acquisition. He says it can only be developed in young children, and discusses the research behind this.

Polylog explains how algorithms can be turned into circuits and encoded as satisfiability problems, which lets you effectively run them backwards. It uses this to motivate breaking RSA, general reductions to SAT, and what P vs NP is really asking.

Benn Jordan's video about storing a PNG image in a living bird, a starling that learns to reproduce encoded sounds. It mixes ornithology, audio engineering, and data encoding.

PBS Space Time episode on the black hole information paradox, covering how black holes seem to destroy information, the no-hair conjecture and the proposed role of the event horizon. It is part of a series on black holes.

Yuta, a Japanese YouTuber, explains whether Japanese speakers can read Chinese text. He shows how shared kanji let people guess meanings while the two languages remain completely different.

A Veritasium video on how a feud between Russian mathematicians led to Markov chains and the modern prediction algorithms built on them. It covers the history and the math behind them.

Tom Scott explains YouTube's 11-character video IDs, which are base-64 encoded 64-bit numbers, and whether YouTube could ever run out of them. It covers counting systems and base 64 briefly.

Tom Scott explains why subtitles often differ from dubbed dialogue, covering how translation, timing and reading speed shape each. It features translators and voice talent working in several languages.