
OpenAI正在扩大其面向学术研究人员的ChatGPT计划,新增5万个资助名额,每个名额最多覆盖三人,现计划覆盖20万名研究人员。新资助名额提供ChatGPT Plus,而非Pro,且按符合条件的申请人先到先得发放,而非抽签。
OpenAI is expanding its ChatGPT for Academic Researchers programme with 50,000 new grants, each covering up to three people, and now aims to reach 200,000 researchers. New grants get ChatGPT Plus instead of Pro, and go to the first eligible applicants rather than by lottery.
OpenAI周一宣布,将向学术研究人员再提供5万个免费ChatGPT资助名额。申请将于11月23日开放,名额按符合条件的申请人先到先得发放,而非抽签。每个资助名额覆盖一名研究人员及最多两名合作者,总计最多覆盖15万人。
OpenAI will give 50,000 more free ChatGPT grants to academic researchers, the company announced on Monday. Applications open on November 23, and grants go to the first eligible applicants instead of by lottery. Each grant covers a researcher and up to two collaborators, or up to 150,000 people in total.
这里有个条件。试点计划中的研究人员获得的是OpenAI的最高级别ChatGPT Pro,而新资助名额提供的是更便宜的Plus计划。OpenAI表示,大多数试点用户的使用量在Plus限额之内,此举能让其覆盖更多团队。
There is a catch. Researchers in the pilot got ChatGPT Pro, OpenAI’s top tier, while new grants come with the cheaper Plus plan. OpenAI says most pilot users would have stayed within Plus limits, and the switch lets it reach more teams.
OpenAI现在计划覆盖20万名研究人员,是该计划7月试点启动时设定目标的两倍。当时,每位研究人员最多可邀请四名合作者。迄今,33个国家的2万名研究人员已加入。
OpenAI now aims to reach 200,000 researchers, double the goal it set when the pilot launched in July. Back then, each researcher could invite up to four collaborators. So far, 20,000 researchers in 33 countries have joined.
OpenAI还发布了一项研究,分析超过1800名科学家在试点中如何使用ChatGPT。平均而言,阅读和学习占其科学信息的28%,写作占27%。受访参与者表示,如果没有AI,同样的工作将花费近两倍的时间。
OpenAI also published a study of how more than 1,800 scientists used ChatGPT in the pilot. On average, reading and learning made up 28% of their science messages, and writing 27%. Surveyed participants said the same work would have taken nearly twice as long without AI.
这一数字来自自我报告,OpenAI表示,该研究并未证明AI能加速发现。它还指出,资金、人员和实验室容量是测试新想法的限制因素。
That figure is self-reported, and OpenAI says the study does not prove AI speeds up discovery. It also names funding, staff and lab capacity as limits on testing new ideas.
这些资助是OpenAI多项科学推动计划之一。它向美国大学实验室提供其生物学模型GPT-Rosalind的免费席位,并向西蒙斯基金会的一项帮助数学家核查AI结果的项目提供了500万美元。8月份,其未发布的Astra模型声称解决了数学领域的十个未解决问题。
The grants are one of several OpenAI science pushes. It offers free seats on GPT-Rosalind, its biology model, to US university labs, and gave $5M to a Simons Foundation effort to help mathematicians check AI results. In August, its unreleased Astra model claimed to have solved ten open problems in maths.
竞争对手也在争取科学家。Anthropic于6月推出了面向实验室的AI工作台Claude Science。Google DeepMind在10月份向200名欧洲研究人员开放了其科学模型。
Rivals are courting scientists too. Anthropic launched Claude Science, an AI workbench for labs, in June. Google DeepMind opened its science models to 200 European researchers in October.