In‑Depth Interview: HKU Vice President and Tencent Advertising Tech Lead Discuss the Future of AI Talent

AI Frontiers and Trustworthy AI

Professor Jin Guoqing explains that AI is already permeating daily life through large‑language and vision models used in autonomous driving, image, text, PPT, music and video generation. He predicts the next frontier will be embodied AI—integrating hardware such as robots with software agents to act as a robot’s "brain"—while emphasizing the need for trustworthy, responsible, and governable AI.

Evolution of Recommendation Systems

Since the first collaborative‑filtering system introduced by Xerox PARC in 1992, recommendation technology has progressed through four generations: (1) collaborative filtering based on similar user preferences; (2) traditional machine‑learning models (logistic regression, GBDT) with heavy feature engineering, exemplified by the Netflix Prize; (3) deep‑learning architectures that established modern recall‑ranking pipelines; and (4) Transformer‑based large recommendation models that continuously scale data, compute and model size.

The authors note that recommendation systems have become core infrastructure for e‑commerce, short‑video feeds and global advertising platforms.

Tencent Advertising AI Innovations

In 2026 Tencent Marketing reported a 20% YoY revenue increase to CNY 382 billion. The technical team pursued a "dual‑technology strategy": expanding model scale and upgrading infrastructure. The RankUp model reached a billion parameters, was deployed across all ad scenarios, and delivered a significant GMV lift in A/B tests. Infrastructure migrated from TensorFlow to native PyTorch with a unified GPU stack. Product suites such as Tencent MiaoSi (an end‑to‑end AIGC creative platform) shortened creative cycles to minutes, while AIM+ reduced manual operation steps by 80% and powered new ad formats (e.g., ninth‑grid ads, instant‑play mini‑games). The mixed‑modal Hy3 model was also integrated into ad‑delivery pipelines.

Algorithm Competition as Talent Platform

The 2026 Tencent Advertising Algorithm Competition (TAAC), now in its seventh edition, attracted participants from over 50 countries and more than 13,000 entrants. It presents real‑world advertising challenges—e.g., unified sequence modeling and feature interaction—providing an open, high‑impact problem set that draws top talent, encourages open‑source data releases (2025 dataset downloaded >10 k times), and bridges academia and industry.

Future of Agent Commercialization

Dr. Zhou Chao outlines three emerging Agent categories for marketing: (1) intelligent‑placement agents that evolve from rule‑based engines to global game‑theoretic optimizers handling audience discovery, slot selection and budget allocation; (2) intelligent‑creative agents that move from one‑shot generation to continuous, millisecond‑level creative iteration driven by real‑time performance feedback; (3) productivity agents (e.g., WorkBuddy) that transform from Q&A tools into full‑featured digital assistants capable of data analysis and strategy recap with a single natural‑language command. The trajectory mirrors a shift from "assistant driving" to "autonomous driving".

Advice for Young AI Professionals

Both speakers stress that knowledge alone is insufficient. Success now requires a blend of curiosity, continuous learning, communication, teamwork, and a resilient attitude. They encourage newcomers to adopt an AI‑native mindset, stay free of legacy technical baggage, and seek roles that offer long‑term growth and stable returns. Passion and curiosity should guide career choices more than short‑term financial incentives.

Conclusion

The discussion returns to the core question: will AI replace humans? The consensus is that AI will automate repetitive, low‑value tasks, but critical judgment, ethical responsibility and strategic decisions must remain human. Embracing AI as a partner—leveraging it for data retrieval, method comparison and workflow efficiency—while retaining final responsibility is presented as the optimal path forward in the rapidly evolving advertising ecosystem.