AI paper by 4 Chinese paved way for ChatGPT, AlphaGo. It's set for greater glory by 2030
The most-cited scientific paper of the 21st century was written by four Chinese researchers in 2016, and it is on track to become the most-cited paper of all time, according to an analysis by British scientific journal Nature.
Advertisement
The paper's findings helped to pave the way for artificial intelligence (AI) tools like ChatGPT and AlphaGo, beating out some of this century's biggest scientific breakthroughs for citation impact.
An analysis released by Nature's news arm on Tuesday examined the 25 papers published since 2000 with the highest number of citations across five academic citation databases, reaching the final list using the median across the databases.
The leading paper received between 103,756 and 254,074 citations, depending on the database consulted. Published at the 2016 IEEE Conference on Computer Vision and Pattern Recognition, the paper deals with ResNets, or deep residual learning networks.
Titled 'Deep residual learning for image recognition', the paper was written by He Kaiming, Ren Shaoqing, Sun Jian and Zhang Xiangyu – all members of Microsoft Research Asia at the time.
Advertisement
When neural networks are made very deep, their performance can worsen because the learning signal gets weaker as it goes through more layers.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
South China Morning Post
3 days ago
- South China Morning Post
AI content detector: why does China dismiss it as ‘superstition tech'?
With the graduation season approaching, many Chinese universities have introduced regulations setting clear requirements for the proportion of artificial intelligence -generated content – or the 'AI rate', as it is called – in theses. Advertisement Some universities have used the AI rate as a deciding factor in whether a thesis is approved. The rule is intended to prevent academic misconduct, as educators have become increasingly concerned about the unregulated use of AI in producing scholarly literature, including data falsification and content fabrication, since the public debut of generative AI models such as ChatGPT However, an official publication of the Ministry of Science and Technology has warned that using AI content detectors to identify AI writing is essentially a form of 'technological superstition' that could cause many unintended side effects. AI detection tools could produce false results, the Science and Technology Daily said in an editorial last Tuesday, adding that some graduates had complained that content clearly written by them was labelled as AI-generated. Advertisement Even a very famous Chinese essay written 100 years ago was evaluated as more than 60 per cent AI-generated, when analysed by these tools, the article said.
South China Morning Post
02-06-2025
- South China Morning Post
Microsoft sees future of the internet as AI agents crawling websites for us, tech willing
Microsoft sees artificial intelligence transforming the internet as fundamentally as mobile phones have over the past two decades. But the technology's limitations could curb its grand vision. Generative AI – which creates content based on a user's request – burst into the zeitgeist in late 2022 when Microsoft-backed OpenAI launched ChatGPT , a conversational chatbot that could take a simple request and generate anything from a limerick to a college essay. Less than three years later, Microsoft has a plan to move beyond ChatGPT and its copycats by creating the foundation for a new version of the internet. Microsoft calls it the 'open agentic web', with users sending AI-powered 'agents' out into the void to do their bidding. Casual consumers primarily interact with AI now through a Google search – one that repeatedly drums up false answers – or a ChatGPT-style chatbot that generates a conversation. Microsoft believes AI agents are the future of the internet. Photo: Reuters In Microsoft's eyes, chatbots are old news.
South China Morning Post
02-06-2025
- South China Morning Post
China boosts research of Southeast Asian populations overlooked in Human Genome Project
For more than two decades, the Human Genome Project (HGP) – a landmark scientific endeavour led by Western nations – mapped humanity's genetic blueprint, yet the rich diversity of Southeast Asia was overlooked. Despite being home to nearly 300 million people, including the world's largest indigenous population, mainland Southeast Asia (MSEA) contributed a mere 1.57 per cent to global genomic databases, with most data derived from diaspora communities rather than local populations. Chinese researchers say they now aim to rectify the omission. Scientists from the Kunming Institute of Zoology, an affiliate of the Chinese Academy of Sciences, have spearheaded a decade-long collaboration with 34 Southeast Asian research teams, culminating in the first comprehensive genomic atlas of the region – the SEA3K data set – which was published in Nature on May 14. 02:43 Nobel Medicine Prize awarded to US duo for 'fundamentally important' discovery of microRNA Nobel Medicine Prize awarded to US duo for 'fundamentally important' discovery of microRNA The study revealed striking insights, including that MSEA populations harbour unique genetic adaptations to tropical challenges as well as distinct Denisovan ancestry linked to Russia's far east. Before China took a role in the research, fewer than 200 indigenous genomes from MSEA existed in global databases, with neighbouring India contributing most via Southeast Asian diaspora samples, according to the paper. Even city state Singapore, a regional sequencing hub, accounted for 92 per cent of the region's limited data, leaving Cambodia, Laos, Myanmar, Thailand and Vietnam virtually invisible in the genomic era. 'For over a decade, we conducted fieldwork in Southeast Asian rainforests, adhering to local ethical protocols, fostering community engagement and documenting indigenous cultural and linguistic contexts,' corresponding author Su Bing said in an interview with China Science Daily on May 19. Researchers collected samples from Southeast Asian populations covering six countries, five prominent language families and 30 ethnic languages. They completed genome sequencing for 3,023 cases, including 40 high-accuracy long-read sequencing data.
