New Era of Discovery: Google Research at I/O 2026
Google/DeepMind
DeepMind
Synaptics
At Google I/O 2026, Google Research unveiled breakthroughs including Gemini for Science, Co-Scientist, and Empirical Research Assistance (ERA) to accelerate scientific discovery. In healthcare, they introduced Symptom AI, AMIE, and MedGemma, while also launching Coral NPU for edge AI and advancing WeatherNext for cyclone prediction.
At Google I/O 2026, Google Research showcased a new agentic era of AI, presenting tools like Gemini for Science, which includes Computational Discovery (built with ERA and AlphaEvolve) and Hypothesis Generation (built with Co-Scientist). ERA is a research coding system that has helped predict hospital admissions and runoff, while Co-Scientier is a multi-agent system tackling challenges like antimicrobial resistance. For healthcare, Symptom AI showed clinicians preferred its differential diagnoses twice as often; AMIE advanced in multimodal medical data, and MedGemma with over 5M downloads enables multimodal medical applications. Coral NPU as an open-source ML accelerator launched the Synaptics Coralboard for edge AI. WeatherNext predicted Hurricane Melissa's intensification five days in advance, and Groundsource used Gemini to create a flash flood dataset. Google also introduced Paper Assistant Tool (PAT) for peer review and Gemini Deep Think for solving open research problems.
- Сокращения
- ERA = Empirical Research Assistance — Эмпирическая исследовательская помощь
- LLM = Large Language Model — Большая языковая модель
- PHR = Personal Health Record — Персональная медицинская запись
- NPU = Neural Processing Unit — Нейронный процессор
- ML = Machine Learning — Машинное обучение
- PAT = Paper Assistant Tool — Инструмент ассистента для статей
Source: Google Research —
original