Xiong Hui: Career Landscape in the AI Era

📺 Video: Click to watch

---

Source: Economist Circle / Sina Finance | 2026-03-23

Hello everyone, I'm Xiong Hui from the Hong Kong University of Science and Technology (Guangzhou). I previously taught at Rutgers University in New Jersey, USA, progressing from Assistant Professor to Associate Professor, Full Professor, and eventually Distinguished Professor. Our field has the privilege of a "revolving door," allowing for two-way exchange between academia and industry. It was precisely because of this that I joined Baidu in 2015 as Vice President of Baidu Research Institute and Chief Scientist, serving until 2020.

I'm very happy today to have the opportunity to share with you how the development of AI in the era of artificial intelligence will impact and disrupt our education system and future talent cultivation.

The Four Realms of Human Intelligence

I categorize human intelligence formed through education into four realms:

1. Encyclopedic Knowledge — Accumulation and memorization of professional knowledge

2. Cross-Domain Transfer — Application across different fields

3. Precise Prediction — Accurate inference and forecasting

4. Creation from Nothing — Innovation from 0 to 1

The emergence of Large Model AI directly surpasses the first two realms. In terms of Encyclopedic Knowledge, humans can no longer compete with AI; in Cross-Domain Transfer, AI leaves humans far behind. Over the past year, AI has also been rapidly catching up in Precise Prediction (inference and forecasting).

Where Should Education Go?

We have entered the era of the Data Flywheel: the data flywheel drives the technology flywheel, which in turn spawns the industry flywheel. Since AI has already surpassed humans in Encyclopedic Knowledge, Cross-Domain Transfer, and even Precise Prediction, what core capabilities remain for humans?

The answer is Creation from Nothing—the ability to innovate from 0 to 1. Current large models based on the Transformer architecture cannot yet achieve this kind of innovation.

Two Precious Human Abilities

In the AI era, two abilities have become particularly precious:

• Questioning Ability: More than 50% of a large model's capabilities cannot be awakened by ordinary users' questions. Only those skilled at asking questions can fully unlock AI's value.
• Appreciation/Judgment Ability: After AI generates content, the ability to filter, judge, and select directly impacts usage effectiveness.

Xiong Hui used his own experience preparing a speech as an example—using AI to draft the speech structure in just a few minutes, but the final three core viewpoints and the taste in selecting cases came from human appreciation and creativity. This entire process embodies Human-AI Integrated Innovation Capability.

Exponential Gap in Learning Power

In the AI era, the gap between people is widening exponentially through learning power. Those who master AI tools see their learning ability grow exponentially, while most others can only achieve linear improvement. 4 hours of AI-assisted learning may surpass over ten hours of effort by others.

The core challenge is that very few people truly master AI tools, especially within the education system, where teachers from primary school to university generally lack relevant awareness.

Xiong Hui cited reading papers as an example: Papers that used to take days to study can now be thoroughly understood in half an hour using large models + precise questioning.

---

Original Link: Xiong Hui: Career Map in the AI Era