Topic
Foundations
Learning, networks, and the ideas underneath modern AI.
7 entries in Super Intelligence
Foundations
Artificial intelligence
Artificial intelligence is the research field, and the set of technologies, that builds machines to perform tasks associated with human intelligence, from pattern recognition and prediction to language use and generation.
Foundations
Transformer
The transformer is a neural network architecture built on self-attention, introduced in 2017, whose parallel structure and scaling behaviour made modern language models possible.
Foundations
Machine learning
Machine learning is the branch of artificial intelligence that fits models to data, allowing computers to classify, predict, and generate without being programmed with explicit rules for each case.
Foundations
RLHF
RLHF trains a model against a learned model of human preference, using pairwise comparisons, a reward model, and reinforcement learning with a penalty that keeps it close to its starting point.
Foundations
Neural network
A neural network is a machine-learning model of layered units that learn weighted connections from data, the basis of modern computer vision, speech recognition, and large language models.
Foundations
Retrieval-augmented generation
Retrieval-augmented generation pairs a language model with a document index, retrieving passages at query time so that answers can draw on sources the model was never trained on.
Foundations
AI benchmarks
An AI benchmark is a fixed set of tasks and a scoring protocol used to compare systems. Benchmarks shape what laboratories optimise, and they are vulnerable to contamination, saturation, and gaming.