I think both have value. Big industry-specific models will be powerful in areas like healthcare or finance, but AI agents look more flexible for the future. Probably we’ll end up with a mix of both.

Both paths are important, but they serve different layers of the AI future.

• Large industry-specific models provide stability, regulation, and domain expertise — they will dominate in critical sectors like healthcare, law, and finance.
• AI agents, however, represent the real “third generation” shift: adaptive, decentralized, and capable of negotiating across systems and cultures.

In my own research (Deconstructing AI Power: From Political Capital to Algorithmic Control), I argue that the real future lies in how we balance these two forces — the institutional power of large models and the flexible agency of autonomous systems. The challenge is political as much as technological.

Preprint Deconstructing AI Power: From Political Capital to Algorithmic Control

Both parties play crucial roles in the future of AI. Industry-specific models offer high accuracy, domain relevance, and better compliance with regulations, making them ideal for sectors like healthcare, law, and finance. Meanwhile, AI agents bring autonomy, adaptability, and the ability to perform complex, multi-step tasks by interacting with tools, APIs, and users in real time. I also suggest the synergy between both to offer powerful and context-sensitive response for real-world deployment.

Dear Fugui Luo, Both, but for different wins. Industry-specific LMs dominate where deep domain accuracy, compliance, and trust matter. AI agents win where orchestration, automation, and user-facing task execution matter. The future is hybrid: domain-tuned models powering intelligent agents.

📩I would be thankful if you gave me your valuable review on my paper:

Article Comparative analysis on AI-driven human digital twin for per...