Google Deepmind was founded in 2010 (as Deepmind) by a founding team of Demis Hassabis, Mustafa Suleyman and Shane Legg.
Deepmind achieved early success by pioneering deep reinforcement learning. The DQN program, which they developed, was applied to play games on the Atari 2600 console and achieved human-level performance in various games without prior knowledge of the game rules.
Google acquired DeepMind in 2014. Its achievements since then include:
- AlphaFold can predict 3D protein structures with remarkable accuracy.
- AlphaGo made history by defeating top-ranked Go players worldwide.
- WaveNet is an advanced text-to-speech system that generates natural-sounding speech.
- RT-2 is a robotics system that learns from web and robotics data to learn generalised instructions for robotics control.
- MedPaLM is a medical-specific chatbot which utilizes large language models to generate safe and helpful answers for clinical applications.