How Should We Study in the Era of Cognitive Revolution?

On April 3, Dr. Yansong Li from the Deqing Alpha Institute of the University of Science and Technology of China and Liii Network Technology Co., Ltd. was invited to give a lecture at Luoyang Normal University's Grand Lecture Hall. The lecture, titled "The Cognitive Revolution in the Era of Large Models: How Should We Learn?", took the audience on a journey through the history of artificial intelligence (AI) development and prompted deep reflection on the profound changes in human learning patterns in the AI era.

Starting with "The Past and Present of AI"

Dr. Li began with a lively overview of AI’s evolution—from 20th-century symbolic and connectionist approaches to today’s reinforcement learning and large-scale models. He noted that early AI’s “rigorous logic” was hampered by “vague definitions,” illustrating the point with humorous “garbage logic” examples that both educated and entertained.

Breakthroughs and Challenges in AI Development

He examined how machine learning and reinforcement learning are reshaping fields from visual recognition to language generation and from gaming to real-world decision-making. He also stated frankly, “No matter how powerful AI becomes, it still depends on human experience and logic”, especially given ongoing challenges in training time and generalization.

Next Stop: Agents and AI Autonomy

When looking into the future of AI, Dr. Li particularly pointed out the development trend of "Agent" systems. He predicted that future AI will have stronger autonomy, capable of independently searching for information, processing documents, and running code, gradually moving from "human scheduling" to "AI autonomous planning." He cited METR research showing that autonomous AI capability follows a "Moore’s Law" of its own: "The time AI can work autonomously doubles every seven months."

How Should Humans Learn? What Can AI Bring?

Dr. Li argued that future AI will be more than a tool—it will be a partner. To tackle continuous learning, ultra-long contexts, and cross-dimensional reasoning, AI still relies on the experience and logic humans provide. He encouraged students to embrace multimodal, interactive learning and to use AI to build their own knowledge networks.

Liii STEM: Making Interactive Learning and Scientific Writing More Efficient

Beyond sparking thoughts, Dr. Li also introduced Liii STEM, the intelligent scientific writing platform he and his team created. This is an AI scientific writing platform specifically designed for scientific writing scenarios in the AI era. The platform's built-in AI interactive writing features will help build personal knowledge graphs. The platform integrates core functions such as code execution, paper typesetting, chart creation, and intelligent generation, along with innovative experiences like "magic paste," dedicated to creating a zero-threshold scientific writing environment and empowering technological innovation and AI education.

He joked: "In the past, scientific writing was like piecing together a puzzle with various tools, but we hope that through AI, you can truly complete the entire scientific writing process in one place, from 'thinking' to 'writing', seamlessly online."

Lively Atmosphere and Continuous Interaction

The session ended with enthusiastic questions on AI ethics and career prospects. Many students said the talk not only delivered cutting-edge insights but also sparked profound reflection on how we learn. Dr. Li closed with a rallying call:

"Don't be afraid of difficulty, leave the difficult tasks to AI, let's focus on more creative things."

This might be the most thought-provoking learning revolution in the era of large models.