AI and the Future of Learning: A New Perspective
A friend once proposed a seemingly naive idea: "co-learning theory." The premise is that if your coding skills are currently inferior to AI, your next four years in college will be spent learning alongside an AI that is constantly evolving, potentially outpacing your own progress. This raises the question: is there a valid logic to this idea, and how should it influence our approach to learning and career planning?
The Rapid Advancement of AI
The pace of AI development is undeniable. In 2022, AI models were being experimented with to create videos. By February 2024, OpenAI released Sora, demonstrating incredibly realistic video generation capabilities, including complex interactions and realistic physics. Similarly, AI's problem-solving abilities have dramatically improved. While basic calculations posed challenges for models like ChatGPT 3.5, DeepMind's Alpha Geometry has since solved a significant number of international math Olympiad problems, even surpassing human performance on certain challenging questions.
The Illusion of Stagnation
Despite these advancements, many still view AI as a mere "toy," useful primarily for tasks like proofreading. Macroeconomic data suggests that labor productivity hasn't seen a significant boost from AI. Traditional search engines still dominate, with AI chat tools representing only a fraction of total online traffic. However, looking back at previous technological revolutions, this apparent stagnation may be misleading.
AI as a General Purpose Technology
AI is evolving into a "general purpose technology," similar to electricity. The impact of such technologies isn't immediate. The introduction of electricity in late 19th-century American factories didn't initially improve productivity. The turning point came with the electric motor, enabling innovations like the conveyor belt. Real productivity gains only occurred after complete reorganization of factory layouts and processes.
For individuals to truly feel the impact of a technology, it requires:
-
Technological Innovation
-
Technology Spread
-
Commercialized Product Development
-
Reorganization of Organizational Structures
Each stage takes time, but today's technological pace is outpacing product development cycles. Just because we haven't seen specific AI applications yet doesn't mean they are not technically feasible or lacking potential.
The Speed Challenge for Developers
The primary challenge for AI developers is not a lack of powerful models, but the rapid pace of development. The speed makes it difficult to assess and adapt. The core issue: technological evolution is far exceeding product development. Companies face the dilemma of potentially being left behind if they hold back current developments.
AI's Expanding Capabilities: Agent Models
Agent-type AI models are being tested in scenarios such as playing Pokemon games. These models can disassemble tasks, execute instructions, check results, and complete objectives without specialized training. This capability translates to common white-collar tasks:
-
Writing emails
-
Creating data tables
-
Filing documents
-
Organizing meeting materials
This suggests that AI is rapidly approaching the point where it can automate entry-level white-collar jobs.
The Investment Boom in AI
Major tech companies are investing unprecedented resources in AI. In 2025, Meta, Amazon, Alphabet, and Microsoft are projected to invest approximately $3.2 trillion in AI and data centers. This is equivalent to the annual GDP of Finland or 15 years of high-speed rail construction in China. Additionally, over 60,000 new AI companies are developing AI products.
AI's Self-Learning Capabilities
Artificial intelligence has been taught to self-study, enabling it to become more powerful without as much human intervention. AI is progressing on a path from simple learning to self-improvement:
- Self-Coordination: AI can strengthen itself through self-play, similar to reinforcement learning.
- Original Learning: AI accumulates learning experience across various tasks, allowing it to learn new skills faster.
- Evolution of Math: Inspired by biological evolution, AI can innovate, create, and modify its algorithms, constantly improving.
- Self-Improvement: AI can actively modify its own core algorithms and structural designs, leading to exponential growth in intelligence.
When AI achieves self-improvement, it could potentially explode, surpassing human intelligence in a very short time.
The Potential Impact of AGI
Human growth is limited by physical constraints. We need sleep, food, and time to learn. AGI, however, does not have these limitations. It can operate 24/7 with millions of "scientists" working simultaneously. This could break our traditional understanding of scientific progress, potentially leading to breakthroughs in areas like dark matter, nuclear fusion, and cancer research.
Planning Your Career in the Age of AI
Given the rapid advancements in AI, individuals should consider how to plan their careers. Is it wise to follow the traditional path of starting with basic tasks and gradually gaining experience? Or are you willing to dedicate a decade to a role that may be obsolete in the future?
Reframing the Job Market
The apparent disappearance of jobs due to AI is a superficial phenomenon. AI increases the output per programmer, reducing employment costs and improving return on investment (ROI). This increased ROI encourages companies to invest more, hire more individuals with AI skills, and expand their projects.
The Changing Landscape of Employment
AI isn't eliminating employment; it's changing its distribution, starting points, and speed. Individuals are now required to understand systems and perform quickly from the outset. This shifts the distribution, starting point, and speed of employment.
AI's Impact on Learning Opportunities
AI may be quietly limiting traditional growth paths. Companies are increasingly seeking designers and system engineers who understand the business and can manage AI effectively, instead of focusing on hiring large numbers of programmers. Newcomers may not have the opportunity to learn from basic tasks and accumulate system intuition, as these tasks are increasingly handled by AI.
Actively Designing Your Growth Path
To address this challenge, individuals must actively design their own growth paths. This involves:
-
Taking Risks: Entering real-world problem situations as early as possible.
-
Leveraging AI: Using AI to expand influence and achieve results quickly, even with imperfect skills.
-
Thinking Like an Entrepreneur: Developing minimal viable products and using AI to assist in project completion.
-
Turning Knowledge into Prompts: Using AI for programming, organization, and testing to make it a project partner.
By proactively adapting to the changing landscape, individuals can thrive in the age of AI.