中国如何在人工智能竞赛中脱颖而出?
此外,对于中国滥用AI技术和侵犯隐私的指责,刘冬梅认为,国际舆论对中国AI发展的误解,部分源自于信息不对称和意识形态偏见。中美两国尤其需要加强沟通交流,以增进相互理解与信任。
人工智能、特别是大模型的发展,取决于算力、算法和数据这三个关键要素。算力与芯片相关,算法与基础能力有关,数据则与语言和用户市场有关。这三个关键要素背后又是人才、巨量资金投入和创新生态问题。
The development of AI, particularly large-scale models, relies on three key factors: computing power, algorithms, and data. Computing power is closely related to chip technology, algorithms depend on foundational capabilities, and data is tied to language and user markets. Behind these key factors lie the issues of talent, massive investment and innovation ecosystems.
简单地说,在这几个方面美国都具有领先于其他国家的优势。
Put simply, the US holds advantages over other countries in these aspects.
在基础大模型方面,中国公司当前的确略微落后于美国,这背后有算力不足的原因,也有高水平人才短缺和投入不足的问题。
In terms of foundational large-scale models, Chinese companies are slightly lagging behind the US. This is due to insufficient computing power, as well as shortage of high-level talent and insufficient investment.
但总的来说,无论是论文发表、专利申请、风险投资的数量,还是大模型的研发,中美都是人工智能发展明显领先的两个主要国家。
However, overall, both China and the US are among the leading countries in the development of AI, be it in terms of number of publications, patent applications, venture capital investment, or the research and development of large-scale models.
中国要想迎头赶上,必须解决以下三个问题:
To catch up in the AI race, China must focus on the following three areas:
一是算力短板,特别是高性能人工智能芯片问题;
First, tackling the deficiency in computing power, especially in high-performance AI chip technology.
二是算法短板,根本上是高水平基础研究人才的培养问题;
Second, addressing the shortfall in algorithms, fundamentally by nurturing top-tier talent in basic research.
三是商业生态营造,即如何充分利用中国巨大的市场需求优势,探索能够商业落地的应用场景,尽快实现商业营利,并在应用中快速迭代发展和反哺基础模型研发,形成人工智能发展的正反馈机制。
Third, fostering a robust commercial ecosystem, which involves leveraging China's vast market demand advantage to explore commercially viable application scenarios, swiftly realizing profitability, and imperatively developing and reinvesting in basic model research through application, thereby creating a positive feedback loop for AI development.