Llion Jones
Llion Jones is a Welsh AI researcher best known as one of the eight co-authors of Attention Is All You Need, the 2017 paper that introduced the Transformer architecture and quietly rewired the future of Large language model research. He studied Artificial Intelligence and Computer Science at the University of Birmingham before joining Google, where the idea for a model built entirely on Self-attention — discarding Recurrence and Convolutions altogether — took shape alongside collaborators Ashish Vaswani, Noam Shazeer, Niki Parmar, and Jakob Uszkoreit.
All eight authors of the paper were credited as equal contributors, with author order randomized, and after the paper, each of the authors left Google to join other companies or to found startups. Jones's path led to Tokyo, where he co-founded Sakana AI and serves as Chief Technology Officer. He has described the venture's philosophy in terms of nature-inspired, swarm-like intelligence: "Sakana is the Japanese word for fish... we're trying to do things like swarm intelligence. So, an individual fish might not be very smart, but a school of fish can maybe have emergent behaviour and we want to apply these kind of nature-inspired ideas to AI research", echoing themes found in Agent-based modeling.
Before the Transformer, sequence tasks like Neural machine translation relied on Recurrent neural network and Long short-term memory architectures, processed step-by-step in a Sequence to sequence learning framework. By replacing recurrence with parallel Attention computation, Jones and his co-authors unlocked the scale that later enabled GPT, GPT-3, BERT, T5, and the Vision transformer — a lineage traceable on the AI and LLM Papers Reading Map alongside descendants like Scaling laws for neural language models and Chinchilla.
Jones remains a quietly influential figure whose work sits at the origin point of the modern Large language model explosion, reflecting on how a small, freedom-filled research team stumbled into one of computing's pivotal architectures.