Routine-biased technological change (RBTC) has profoundly restructured the skill premium structure of the labor market, drastically compressed the living space of routine jobs, and thus exposed the structural lag of the traditional college employment guidance system in implicit competency mapping. This paper aims to deconstruct the dual mechanism of large language models (LLMs) in the micro-level employment intervention process. By comparing with the traditional static tutoring model, the study points out that LLMs, relying on their unstructured feature extraction capability, have effectively broken through the job search information barriers of job seekers and realized the paradigm shift from "typesetting polishing" to "dynamic semantic mapping". However, in practical application scenarios, the underlying alignment mechanism and text generation logic of algorithms can easily strip job seekers of their real practical baselines, inducing the risks of "ability falsification" and Akerlof's adverse selection in the labor market. To this end, this paper proposes that a high-fidelity dynamic game confrontation mechanism must be introduced into the intelligent empowerment framework, and the negative backlash of text "overstepping reconstruction" should be blocked through cognitive load limit testing, so as to provide a theoretical framework for the development of intelligent employment assistance systems with anti-reverse dependence attributes.

Sheng Yang. Mitigation of College Students' Employment Friction and Adverse Selection Risk under the Intervention of Large Language Models: A Mechanism Analysis from the Perspective of Routine-Biased Technological Change. Frontiers in Educational Research (2026), Vol. 9, Issue 4: 154-157. https://doi.org/10.25236/FER.2026.090423.

[1] Zhou, W.X., Li, S.Y. and Feng, Y. (2022) A Review of College Students' Employment and Their Employment Dilemma. China University Students Career Guide, 7, 3-8.

[2] Chen, S., Han, Q. and Zhang, B.C. (2025) The Level of AI Technology in Enterprises and Changes in Labor Demand Structure: New Methods and Findings Based on Large Language Models. Journal of Technology Economics, 44(1), 1-13.

[3] Zhao, Z.G. (2024) The Impact of Artificial Intelligence on College Students' Employment and Optimization Countermeasures. Sichuan Labor Security, 4, 76-77.

[4] Long, Y.T., Liu, H.B. and Cai, Y.Z. (2020) The Impact of Artificial Intelligence Technology on Labor Employment: A Literature Review Perspective. China Soft Science, 12, 56-64.