The Paper | Fan Ling: AI is currently mostly "talking", next it needs to "work"

AI Enters a Shining Moment: Xuhui's 'Big Dipper'

On April 29, General Secretary Xi Jinping inspected the 'Model Speed Space' in Shanghai. Xuhui District has formed an AI ecological layout of 'Big Dipper Formation' + 'Stars Shining'. Special Zhan Technology, as one of the benchmark enterprises of the 'Big Dipper', is a unicorn company focused on 'content + artificial intelligence', dedicated to building an enterprise-level content AI platform.

From 'Left Brain' to 'Right Brain': Unlocking AI's Creativity

Fan Ling, the founder of Special Zhan, holds a PhD from Harvard and previously taught at Berkeley. He proposed that humans have a left brain (logical, stronger) and a right brain (humanities, arts). Currently, most of AI's efforts are focused on 'left brain' development, while Special Zhan is committed to exploring AI's 'right brain' potential—bringing technology into the creative field to combine content creativity with technological innovation to unleash productivity.

Future Trends: AI Needs to Start 'Working'

Current AI is still 'talking', not 'working'. In the next 3 to 5 years, computing power will be used more for problem-solving. If it was previously focused on Training, it may shift to Reasoning.— Fan Ling

Fan Ling believes that multimodal large models refer to those that can 'call various tools'. Whoever can successfully enable efficient collaboration between humans and tools will have a continuous stream of business opportunities. AI will inevitably replace some 'existing jobs', but it will also create more 'incremental jobs'.

Thoughts on Data Sovereignty and 'Knowledge Oligarchs'

Fan Ling expressed concern about the risks of AI governance. He has shifted from being a believer in 'public cloud' to focusing on data asset security: how to protect the private data of enterprises and individuals? How to avoid large models forming 'knowledge oligarchs' or 'information oligarchs'? In the short term, Special Zhan will strengthen its technical capabilities to help enterprises truly transform unstructured data into private data assets.