Ex-ByteDance/Meituan Recruiter: 100M yuan LLM talent not pricey, apps need 10 staff
In a LatePost interview, Wei Xiaokang, former head of recruiting at ByteDance and Meituan, said top LLM talent at 100 million yuan annual salaries is not expensive, while most AI talent is now oversupplied and AI application companies need only about 10 people. He predicted that white-collar jobs will shrink sharply and that opportunities will concentrate on a few exceptional individuals.
Coverage timeline
晚点 LatePost高洪浩
在大模型竞争中，人才和算力是两种昂贵而稀缺的资源。对算力受限的中国模型公司而言，追赶世界领先水平的希望，很大程度上寄托于那些最聪明的大脑。 2024 年，一个博士毕业生只要研究方向与 AI 有关，就可能获得数百万元年薪。这也是人才泡沫的起点：许多公司花了很多钱，却招来一群根本做不出大模型的人。第二年，腾讯和字节跳动进一步推高了人才价格，极少数顶尖人才的薪酬包达到亿元级别。 但到了 2026 年，市场发生了逆转。随着大模型竞争逐渐收敛到少数公司，人才需求迅速减少。一些有顶级论文、头部模型厂实习经历的优秀毕业生，甚至拿不到面试机会。 最近，我们与 AI 创业者、前字节跳动和美团招聘负责人魏小康对话，回顾这场人才争夺战的起落。 魏小康亲历了中国互联网行业两轮激烈的人才争夺，也是一个观点激进、人才观鲜明的人。 2017 年，他参与搭建字节跳动的招聘体系。那正是中国互联网公司通过高薪、大规模招聘和不断扩大组织，建立竞争优势的时代。字节后来集结了数千名 HR，用超高薪酬在全行业争夺优秀人才。此后他又在美团参与规模化校招，高峰时期一年招聘五六千名毕业生。 但今天，魏小康相信的几乎是一套与过去完全相反的逻辑。 他一方面认为，顶尖天才没被严重低估，“给大模型人才开 1 亿年薪完全不贵”；另一方面，又认为绝大多数优秀人才在基座模型领域已经过剩，未来甚至没有机会入行。 对于 AI 应用公司，他也不理解为什么还需要二十多个产研人员，“十几个善用 AI 的人，就可以完成过去两百人的工作。” 这种判断不仅限于科技行业。他认为，未来几年，大量白领岗位可能消失；互联网大厂这一轮裁员之后，人员规模不会再恢复；过去一百个人进入一个行业，最后有十个人脱颖而出，而未来，可能只有那十个人有机会进入。 这些看似激进的判断，背后有一套一致的逻辑：AI 正在让顶尖人才与普通人才的经济价值进一步分化，组织对人的需求急剧减少，最终绝大多数机会和回报可能集中到极少数顶尖人才手中。 如果说，过去二十年，中国互联网公司的竞争，是谁能够聚集、组织更多聪明人，那么在魏小康设想的 AI 时代，竞争将变成谁能找到极少数天才、人才，并让他们摆脱对大量其他人的依赖。 一个曾经帮助互联网公司大规模争夺人才的招聘负责人，如今相信，AI 将终结支撑互联网行业过去二十年发展的那套人才组织方式。 这是魏小康对未来的判断。它是否会成为现实，仍有许
Related stories
Chinese AI startups HyperVolve and Apexin announce funding and model releases
Chinese AI startup HyperVolve announced tens of millions RMB in angel funding led by Lanchi Ventures, with participation from SenseTime Guoxiang Capital, Lenovo Star, Inno Fund, and Tsinghua alumni seed fund, to build research foundation models using recursive self-improvement. Separately, Apexin, founded by 27-year-old Tsinghua assistant professor Chen Yongchao, released Apex-35B-V1.0, a 35B model claiming SOTA results on 21 benchmarks against comparable models.
2 reports · 1 sources
LeCun paper argues next-token prediction inefficient, proposes JEPA world models
A new paper by Yann LeCun and colleagues mathematically argues that next-token prediction is inefficient and proposes an alternative based on world models and JEPA (Joint Embedding Predictive Architecture). The paper claims a 109M-parameter model outperforms a 72B LLM, and formalizes the Platonic Representation Hypothesis into a theorem with actionable sufficient conditions.
AI Talent War Cools in China: Top PhDs Face Fewer Offers, Tighter Hiring
Chinese AI labs are cooling their talent war, with top PhD candidates facing fewer opportunities and stricter hiring criteria. A 2027 PhD with six top papers and two major internships applied to over 20 companies but received only five or six offers by late October, with some processes ending prematurely. Hiring now focuses on specific gaps, making it harder for even experienced researchers to switch core teams.
1 reports · 1 sources
Microsoft Offers Buyouts to 60,000 Staff as AI Spending Rises
The New York Times reports that Microsoft employees are reacting to the company's buyout offer to 60,000 staff, which comes amid growing AI spending. Employees rushed to an internal site within an hour of the announcement to check if they were eligible. The move marks a shift for a company long known for its stability in the industry.
1 reports · 1 sources
News updates every hour. Subscribe in your RSS reader to get new stories automatically.
Subscribe