AI in 2021
6 events, oldest first. Each links to a page with details and sources.
- OpenAI unveils DALL·E and CLIP ★★★★
DALL·E generated images from text prompts using a 12B-parameter Transformer, and CLIP learned joint image–text representations from 400M image-caption pairs; CLIP became a key component of later diffusion image generators.
- Adam Zsolt Wagner uses reinforcement learning to find counterexamples to open graph-theory conjectures ★★
Wagner's 'Constructions in combinatorics via neural networks' (arXiv 2104.14516) used a simple cross-entropy RL method to find explicit counterexamples to several published conjectures in extremal combinatorics and spectral graph theory.
- Anthropic launches with a focus on AI safety ★★★
Anthropic, founded by former OpenAI researchers including Dario and Daniela Amodei, announced a $124M Series A to build reliable, interpretable and steerable AI systems.
- GitHub Copilot and OpenAI Codex bring LLMs to programming ★★★★
GitHub launched Copilot as a technical preview, an AI pair programmer powered by OpenAI Codex, a GPT model fine-tuned on public code; the Codex paper introduced the HumanEval benchmark.
- NASA's ExoMiner deep-learning model validates 301 new exoplanets from Kepler data ★★
NASA's ExoMiner neural network statistically validated 301 Kepler planet candidates as real planets in one batch, bringing the validated count to 4,569 (Astrophysical Journal, 2021).
- DeepMind and mathematicians use machine learning to guide new theorems in knot theory and representation theory ★★★
Davies et al. (Nature, Dec 2021) used supervised learning plus attribution to point mathematicians to hidden relationships. That led to a new theorem linking the knot signature to hyperbolic geometry, and to progress on the combinatorial invariance conjecture for Kazhdan–Lusztig polynomials.