近日，谷歌DeepMind的大模型在解决数学难题方面取得了重大突破，这一成果再次登上了国际顶级学术期刊《自然》(Nature)的封面。这项名为FunSearch的技术，利用大模型解决了长期存在的科学难题，产生了以前不存在的可验证且有价值的新信息。DeepMind负责人表示，他们将大模型视为一种创造力引擎，这是第一次有人证明基于大模型的系统可以超越数学家和计算机科学家的认知。
据悉，FunSearch技术的核心是函数(Function)一词的简写，它利用大模型解决了一个困扰数学家60多年的问题。这一问题的解法超出了人类已有的认知水平，而FunSearch则为我们提供了一种全新的解决方案。在Nature论文配套的新闻解读中，DeepMind负责人指出：“我们使用大模型的方式是当做创造力引擎。这是第一次有人证明基于大模型的系统可以超越数学家和计算机科学家的认知。它不仅新颖，而且比当今存在的任何其他东西都更有效。”
这一突破性成果对于数学界和计算机科学界来说具有重要的意义。首先，它证明了大模型在解决复杂问题方面的潜力，为相关领域的研究提供了新的思路和方法。其次，这一成果也为人工智能领域带来了新的启示，有助于推动AI技术的进一步发展。
目前，DeepMind正在与全球多家科研机构和高校合作，将这一研究成果应用于更多领域，如生物学、物理学等。相信在未来，我们将会看到更多基于大模型的创新应用，为人类社会带来更多的便利和价值。
英语如下：
Title: DeepMind’s Large Model Breaks Through 60-Year Mathematical Problem, Innovation Method Surpasses Cognition!
Keywords: DeepMind, Mathematical Problem, Large Model
Recently, Google’s DeepMind large model has made a significant breakthrough in solving mathematical problems, once again appearing on the cover of the world’s top academic journal Nature. This technology, called FunSearch, uses large models to solve long-standing scientific problems and generates new, verifiable and valuable information that did not exist before. DeepMind officials said that they view large models as a creative engine, and this is the first time anyone has proven that a system based on large models can surpass the cognitive abilities of mathematicians and computer scientists.
It is reported that the core of FunSearch technology is the abbreviation of the word “function,” which uses large models to solve a problem that has plagued mathematicians for more than 60 years. The solution to this problem goes beyond human cognitive levels, while FunSearch provides us with a brand-new approach. In the news commentary accompanying the Nature paper, DeepMind officials pointed out, “We use large models as a way to create creativity. This is the first time someone has proven that a system based on large models can surpass the cognitive abilities of mathematicians and computer scientists. It’s not only novel but also more effective than anything else that exists today.”
This breakthrough is of great significance to both the mathematics and computer science communities. First, it proves the potential of large models in solving complex problems, providing new ideas and methods for related fields of research. Second, this achievement also brings new insights into the field of artificial intelligence, helping to promote the further development of AI technology.
At present, DeepMind is cooperating with multiple research institutions and universities around the world to apply this research result to more fields, such as biology, physics, etc. It is believed that in the future, we will see more innovative applications based on large models, bringing more convenience and value to human society.
【来源】https://www.qbitai.com/2023/12/106559.html
Views: 6