Google 研究科学家 John Platt 谈 AI for Science：从凝云减碳到疫情预测
Google 老将 John Platt 聊了 AI 怎么用于科研：减飞机尾迹、FireSat 卫星测火，还有用 Gemini 加蒙特卡洛树搜索拿下 CDC 疫情预测基准第一。
Google 研究科学家 John Platt 在 Latent Space 播客中介绍其团队用 AI 减少飞机凝结尾迹的工作，凝结尾迹约占人为暖化的 1%。他还介绍了 FireSat 卫星用于野火探测的进展。在疾病预测方面，其研究团队的 Project IRA 结合 Gemini 与蒙特卡洛树搜索，在 CDC 基准的新冠和流感一周前瞻预测中获得最高评分。
How does an Academy Award winner with two named asteroids build AI for science?
In the latest @latentspacepod episode, Google’s John Platt discusses:
• Fighting climate change with AI, from airplane contrail reduction (those little white lines behind planes that contribute ~1% of human-caused warming) to fire detection with FireSat satellites.
• Google helping pioneer the application of data to track disease progression, and how that journey has come full circle: our research team's Project IRA (using Gemini and Monte Carlo Tree Search) achieved top marks in recent CDC benchmarks for forecasting COVID and flu cases a week in advance.
• How the field of AI in science has evolved and what lessons he's learned.
• His advice for young scientists, including his own son.
Listen to the full episode: https://t.co/JEbp9h8OZZ
