The World Artificial Intelligence Conference (WAIC 2026) hosted the \u2018AI for Science: Beyond the Concept\u2019 media roundtable in Shanghai, convened by Huawei. Four of China\u2019s leading research institutions \u2013 BAAI, CAS, SLAI and Tsinghua University \u2013 presented their latest AI-driven research outcomes spanning neuroscience, interdisciplinary research platforms, intelligent instrumentation and life sciences. The presentations ranged from sample preparation to data analysis, from literature review to experimental validation, emphasising R&D cycles are shortening from years to days.<\/p>\n\n\n\n
While global R&D investment continues climbing, the rate of scientific discovery is evolving at a slower pace. Experimental cycles still take too long to run, interdisciplinary collaboration faces persistent barriers and research workflows remain fragmented. Across the roundtable, the panelists were clear that AI is there to augment scientists, not replace them – the goal is to relieve researchers of repetitive tasks so they can focus on the insight and judgment only they can provide. The outcomes unveiled at the roundtable tackle this problem from multiple angles.<\/p>\n\n\n\n
Unifying neural data to speed up cross-individual analysis<\/strong><\/p>\n\n\n\n Considered to be the world’s first multimodal foundation model, BAAI’s Wujie\u00b7Brain\u03bc1.0, is built specifically for neuroscience, capable of unifying EEG readings, calcium imaging and neural probe signals within one shared encoding framework. These different types of neural data have historically been hard to reconcile, but the model allows them to be aligned and interpreted through a common architecture rather than treated as incompatible streams.<\/p>\n\n\n\n Its practical value became clear in June 2026, when a study built on the model appeared in Science<\/em>, showing for the first time that reactivating memories has opposite effects on sleep depending on their emotional tone: positive memories were linked to better sleep quality, while negative ones deepened sleep fragmentation. The discovery points towards possible new treatment routes for sleep disorders tied to depression and anxiety. Brain\u03bc has been trained on data from more than 70,000 nights of sleep and has been running continuous, automated analysis across partner laboratories for over a year.<\/p>\n\n\n\n Lei Bo, a researcher at BAAI, pointed out that neuroscience still lacks standardised data and that AI in the field is at an early stage, yet researchers have responded with real enthusiasm. He argued this is a healthier path than the early days of LLM development, since here it’s demand is pulling capability forward, with practical use cases shaping how the models evolve rather than the other way around.<\/p>\n\n\n\n One model, eight disciplines<\/strong><\/p>\n\n\n\n CAS introduced ScienceOne Omni, spanning mathematics, physics, materials science and astronomy. The model rests on a three-layer structure \u2013 Unified Scientific Data Encoding, Real-World Knowledge Alignment and Domain-Specific Task Decoding \u2013 drawing on 170 million scientific papers and equipped with more than 8,000 specialised research tools and skill libraries. It’s a single system capable of cross-disciplinary data interpretation, scientific reasoning and content generation, closing the gap that has long separated narrow, high-performing specialist models from broader generalist ones that lack real domain expertise.<\/p>\n\n\n\n Xu Nan, a researcher at the CAS Institute of Automation, described ScienceOne Omni as far more than an incremental update, referring to it as a reimagining of what scientific foundation models can be \u2013 models built to reason like scientists. In practice, it has reduced literature reviews from weeks to just 20 minutes, lifted report-writing efficiency five to tenfold and is already running across more than 100 research applications throughout CAS.<\/p>\n\n\n\n Full instrument automation, freeing scientists for scientific judgment<\/strong><\/p>\n\n\n\n SLAI, working in partnership with Suzhou National Laboratory, unveiled Owl\u00b7AuraID, a multi-agent system that automates the complete experimental process end to end, from preparing samples through to analysing the resulting data. Rather than tapping into instrument APIs, its agents work the interfaces the way a human scientist would \u2013 watching the screen, clicking through and reading off the data \u2013 closing the gap between software agents, embodied scientific agents and the instruments themselves. The system now works across six categories of precision instruments, cutting crystal structure analysis workload by 50.6%, bringing morphological analysis time down from nine minutes to seven and a half and pushing AI’s autonomous completion rate up from 33% to 80%.<\/p>\n\n\n\n Ouyang Wanli, Vice Dean of SLAI, observed that scientific characterisation calls for deep expertise and careful coordination across instruments, adding that AI now let\u2019s devices connect, work together and optimise themselves \u2013 freeing scientists from hands-on operations so they can focus on scientific insight.<\/p>\n\n\n\n AI evolves from a \u2018support tool\u2019 to \u2018research Infrastructure\u2019<\/strong><\/p>\n\n\n\n During the roundtable, Tsinghua University’s Professor, Yu Li, remarked that AI is shifting from a mere supporting role to becoming a core piece of research infrastructure \u2013 the goal isn’t to replace scientists, he noted, but to lift the burden of repetitive work so they can devote more energy to insight and creative thinking.<\/p>\n\n\n\n From standalone instruments to fully automated pipelines, from models built for one discipline to systems that collaborate across many, from neuroscience through to the life sciences \u2014 AI for Science is visibly transforming every step of the discovery process. As hypothesis generation, experimental validation, data analysis and instrument operation all speed up under AI, the very limits of scientific discovery are being pushed outwards.<\/p>\n\n\n\n FAQs<\/strong><\/p>\n\n\n\n What is the multimodal neuroscience foundation model, Brain\u03bc?<\/strong><\/p>\n\n\n\n Brain\u03bc is the world’s first neuroscience foundation model capable of understanding multiple types of brain signals simultaneously. It unifies EEG, calcium imaging and neural probe data into a single encoding framework, enabling scientists to analyse cross-individual, cross-scenario neuroscience data within a shared architecture. A study it supported was published in Science<\/em> in June 2026.<\/p>\n\n\n\n Why is ScienceOne Omni important?<\/strong><\/p>\n\n\n\n Previously, AI in research followed two paths: generalist models that lack domain depth and specialist models that excel at single tasks but cannot generalise. ScienceOne Omni bridges this gap through its three-layer architecture \u2013 Unified Data Encoding, World Knowledge Alignment and Task-Specific Decoding \u2013 enabling a single model to achieve both broad interdisciplinary understanding and domain-specific depth.<\/p>\n\n\n\n What are the application scenarios for Owl\u00b7AuraID?<\/strong><\/p>\n\n\n\n Owl\u00b7AuraID has been deployed across six types of precision instruments in materials science, chemistry, biology and other disciplines. Researchers can remotely issue instructions to complete the full workflow, from sample placement and parameter configuration to intelligent analysis, enabling 24\/7 instrument operation. It has been validated in three task categories: crystal structure analysis, morphological analysis and internal structure scanning.<\/p>\n","protected":false},"excerpt":{"rendered":" The World Artificial Intelligence Conference (WAIC 2026) hosted the \u2018AI for Science: Beyond the Concept\u2019 media roundtable in Shanghai, convened by Huawei. Four of China\u2019s leading research institutions \u2013 BAAI, CAS, SLAI and Tsinghua University \u2013 presented their latest AI-driven research outcomes spanning neuroscience, interdisciplinary research platforms, intelligent instrumentation and life sciences. The presentations ranged […]<\/p>\n","protected":false},"author":21,"featured_media":55898,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[7027,1463,4664,56,16,25,11958,9862,11960,42,44],"tags":[11477,1108,11890,11891,11892,217,110,11895,11897,5172,5885,11898,1391,11893,11894,11889,11888,11896],"class_list":["post-55897","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-intelligent-technologies-ai","category-apac","category-intelligent-technologies-data","category-east-asia","category-industry-verticals-education","category-industry-verticals-healthcare","category-hh-huawei-hub","category-hh-newsletter","category-hh-top-story","category-industry-verticals-telecom","category-top-stories","tag-academia","tag-ai","tag-ai-for-science-beyond-the-concept","tag-baai","tag-cas","tag-data","tag-huawei","tag-life-sciences","tag-owlauraid","tag-rd","tag-science","tag-scienceone-omni","tag-shanghai","tag-slai","tag-tsinghua-university","tag-waic-2026","tag-world-artificial-intelligence-conference","tag-wujiebrain1-0"],"acf":[],"publishpress_future_workflow_manual_trigger":{"enabledWorkflows":[]},"_links":{"self":[{"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/posts\/55897","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/users\/21"}],"replies":[{"embeddable":true,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/comments?post=55897"}],"version-history":[{"count":7,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/posts\/55897\/revisions"}],"predecessor-version":[{"id":56014,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/posts\/55897\/revisions\/56014"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/media\/55898"}],"wp:attachment":[{"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/media?parent=55897"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/categories?post=55897"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.intelligentcio.com\/apac\/wp-json\/wp\/v2\/tags?post=55897"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}