Posts

A short introduction to a new DeepMind report by Shane Legg and colleagues, which asks how AI might keep developing once it reaches human-level intelligence.

On November 30, 2022, OpenAI released ChatGPT. In hindsight, that day may mark the beginning of the Fourth Industrial Revolution.

A timeline of AI from the 1956 Dartmouth workshop to ChatGPT: symbolic systems, the rises and falls of neural networks, deep learning, and the LLM era.

UIE: a unified text-to-structure framework that models diverse information-extraction tasks and generates targeted structures on demand.

Vector search engines or vector databases are a core piece of infrastracture that fuels every big deep learning deployment in industry.

Amazon Alexa AI's paper "Language Model Is All You Need" reframes natural-language understanding as a question-answering problem.

Notes on a wide-ranging thread about meaning, semantics, and what language models actually learn — from entailment to co-reference.

Filtered Space-Saving (FSS): a compact algorithm for estimating the top-k most frequent items in a data stream using constant memory.

Kurt Gödel's 1931 incompleteness theorems expose the built-in limits of any formal system rich enough to express basic arithmetic.

Weighted random sampling done right with the Alias Method: O(n) setup, O(1) per draw, O(n) memory — with a walk-through of why it works.

A small Python implementation of the logistic map — showing how one simple equation slides from order into chaos.

How Linux really implements threads and processes — and why the kernel treats both as tasks (light-weight processes) under the hood.