As the technological juggernaut marches forward, we stand at an unprecedented historical juncture.
On June 4, 2025, on the main stage of London's SXSW conference, Demis Hassabis, CEO of Google DeepMind and Nobel laureate, captured the audience's attention with a resonant statement: "AI is replacing some jobs, but it will also create entirely new and highly valuable ones."
This is not mere consolation but a precise timetable. The creator of AlphaFold and a pioneer in the field of artificial intelligence used his unique perspective to sketch a new world of work that is on the horizon.
Yet, this talk was far more than another ode to technological optimism.
A deeper analysis of Hassabis's points, combined with the latest employment data and industry trends, reveals a much more complex and profound reality: what we are witnessing is not a simple replacement of jobs, but a fundamental reconfiguration of the entire labor market.
The scale and depth of this transformation may well exceed our imaginations.
The Cold Reality Behind the Data: 83 Million Jobs Lost and Reborn
Let us start with the most direct figures.
The World Economic Forum's "Future of Jobs Report 2023" presents a stunning projection: by 2027, AI will replace 83 million jobs worldwide while creating 69 million new positions, resulting in a net loss of 14 million jobs.
This is not a distant prediction but a reality unfolding right now.
The "Work Trend Index 2025," jointly released by LinkedIn and Microsoft, shows that 75% of knowledge-based roles globally now incorporate generative AI tools. This means that AI is no longer a technology confined to a distant laboratory—it has taken up residence as a “new colleague” at your office desk.
But these numbers tell us much more than that.
A closer look at the net loss of 14 million jobs reveals a counterintuitive phenomenon: the first to be affected are not the traditional blue-collar jobs, but the seemingly "safest" knowledge-intensive roles.
A recent McKinsey study from 2024 found that high-paying jobs, with salaries exceeding $200,000, are more susceptible to AI impact. This is because current AI technology excels at tasks involving information analysis, content generation, and data processing, which are the core components of high-paying white-collar roles.
PwC’s "AI Employment Barometer 2024," analyzing 500 million job postings across 15 countries, found a subtle trend: the wage premium for professionals with AI skills soared from 25% in 2023 to 56% in 2024, nearly doubling.
This figure carries profound implications—the market is revaluing human capital at an unprecedented pace, and AI skills have become the "digital gold" of the new era.
Three New Species: Hassabis’s Vision of Future Work
During his London talk, Hassabis went beyond abstract concepts. He described three emerging forms of high-value work, which I refer to as the "three new species" of workplace evolution:
1. AI Collaborators – Not Just Tool Users, But Commanders of Intelligence
"I would love an AI product that can automatically handle my emails. I'm willing to pay thousands of dollars a month so I don't have to be tormented by endless emails." Although Hassabis’s remark might sound like a joke, it reveals a crucial shift: AI is no longer merely a tool to answer questions—it is an intelligent agent capable of completing entire tasks.
The essence of an AI collaborator’s role is "intelligent orchestration": they need not code themselves but must know how to instruct AI to do so; they do not need to master every technical detail of each model but must understand which tool fits which scenario; they don't have to manage servers, yet they must be adept at issuing clear instructions and verifying outcomes.
This role is rapidly emerging across various industries. In marketing, specialists are responsible for translating product promotion needs into executable AI task flows; in the legal field, AI collaborators assist lawyers in contract review, legal research, and drafting documents; in finance, they design AI-driven processes for investment analysis and risk assessment.
Intriguingly, this role subverts traditional career paths. It does not demand deep programming expertise but requires strong logical thinking; it does not require specific academic backgrounds but demands the ability to learn and adapt rapidly; most importantly, it calls for a new "human-AI collaborative mindset"—one that understands human needs as well as the boundaries of AI capabilities.
2. Scientific Collaborators – Exploring the Unknown Alongside AI
Hassabis mentioned a startling fact in his talk: DeepMind’s AlphaTensor independently discovered the first optimization in matrix multiplication in 50 years, without human intervention. "AI has moved from simulating science to generating science."
What does that mean? Traditionally, scientific research followed the model of "humans form hypotheses and AI helps to verify them." Now, the paradigm has shifted to "AI independently explores while humans guide and assess." The role of the scientific collaborator is to serve as AI's "research partner" in this new model.
In the AlphaFold project, the research team used AI to predict over 200 million protein structures—more than what humanity had amassed over decades. Crucially, this was not AI’s work alone; it was the result of deep collaboration between human scientists and AI. Scientific collaborators need to possess three core competencies:
- Domain Insight: While they need not be top experts, they must have enough knowledge to identify valuable research directions and assess the significance of results.
- AI Coordination: They must understand the strengths and weaknesses of different AI tools and design experiments that maximize AI’s potential.
- Critical Thinking: Since AI can sometimes produce seemingly plausible but erroneous results, scientific collaborators must be adept at detecting and correcting these mistakes.
This mode of work is moving from laboratories to industry.
In drug discovery, scientific collaborators help pharmaceutical companies accelerate compound discovery and design clinical trials using AI; in materials science, they partner with AI to explore new superconductors and quantum materials; and in climate science, they use AI to build more accurate climate models to predict and address environmental changes.
3. AI Ethics Officers – Setting Boundaries in the Age of Intelligence
When asked about regulating AI, Hassabis pointed out a fundamental paradox: "We cannot wait for problems to arise before we set rules; AI is advancing too rapidly."
Behind this statement lies an urgent need for a novel profession: the AI ethics and safety officer.
The importance of this role is evident—AI is neither an atomic bomb nor a microwave but a new category of technology that does not fit into existing regulatory frameworks. The job of an AI ethics officer is "preventative governance": to identify potential issues prior to AI deployment, design safety boundaries within application scenarios, and provide solutions when controversies arise.
Within DeepMind, a professional "red team" mechanism has been established: every AI model must undergo rigorous testing by a specially trained team tasked with provoking biased, erroneous, or harmful outputs before release. This role calls not for traditional legal or ethical knowledge but a new "AI safety mindset":
- Technical Understanding: While they do not develop AI systems, they must grasp how AI works and identify potential failure modes.
- Scenario Imagination: They must foresee risks and social impacts that AI may generate in different contexts.
- Rule-Design Ability: They need to convert abstract ethical principles into specific technical constraints and operational guidelines.
With AI's expansive application in sensitive fields like healthcare, finance, education, and justice, the demand for AI ethics officers is exploding. Tech giants such as ByteDance, Alibaba, and Baidu are heavily recruiting for positions like "model safety test engineer," "AI ethics analyst," and "AI red team expert."
A Redefined Skill Set: The Renaissance of STEM Education and the Upgrade of Humanities
Hassabis made a clear recommendation: if he were to choose a college major today, he would still pick a STEM discipline (Science, Technology, Engineering, Mathematics). However, he also stressed that modern talent needs more than just technical skills—it requires a deeper "AI hybrid capability."
New Dimensions of Technical Fundamentals
Traditional technical education focuses on mastering specific skills—learning programming languages, data analysis techniques, and algorithm principles. In the AI era, however, the essence of a solid technical foundation has fundamentally changed:
- Mathematical Thinking: Not for computation alone, but to understand the logic and limitations of AI models. Knowing the mathematical principles behind an AI model helps in judging when it is reliable and when it might fail.
- Computational Thinking: Not simply about writing code, but about designing task flows that AI can execute. This calls for a new way of thinking—breaking complex problems into manageable steps that AI can comprehend and perform.
- Systems Thinking: Recognizing that AI is not an isolated tool but a component of a broader technological ecosystem. This includes data flow, model integration, and human-machine interaction, among other aspects.
The Tangible Value of Soft Skills
More intriguingly, in an era of advanced AI, seemingly "soft" skills have become even more crucial:
- The Art of Questioning: AI excels at answering questions but struggles to ask the right ones. In an age of information overload, knowing what to ask can be more valuable than knowing the answer.
- Judgment: AI can generate vast amounts of content but cannot discern which content is truly valuable. This requires human experience, intuition, and values to make appropriate choices.
- Connectivity: While AI is skilled at addressing issues within a single field, cross-disciplinary integration and innovation still demand human insight.
- Communication: As AI becomes an integral component of the workplace, the ability to communicate effectively with AI—and to explain its results to others—has become an essential skill.
Industry Divergence: Winners and Laggards
AI's impact varies significantly across industries, reshaping entire economic structures.
Knowledge-Intensive Industries: Greatest Impact, Greatest Opportunity
Professional services such as law, consulting, and accounting are most immediately affected because the core tasks—information analysis, document processing, and expert judgment—are precisely what AI excels at.
At the same time, these industries reap the greatest opportunities for efficiency gains.
Law firms have begun using AI for contract reviews and legal research; consulting firms use AI for data analysis and report generation; and accounting firms employ AI to handle financial audits and tax filings.
The key is that professionals in these fields must learn to collaborate with AI or risk being replaced by colleagues who possess AI skills.
Finance: Balancing Risks and Opportunities
Finance is among the industries with the most widespread AI applications—from algorithmic trading and risk management to customer service and investment advice, AI is transforming every facet of financial services. Yet, finance faces unique challenges: strict regulatory requirements, high costs of error, and the difficulty of establishing client trust.
This means that financial institutions are seeking AI talent with both technical prowess and deep industry expertise combined with a strong sense of risk management.
As a result, the salaries for AI experts in finance remain high, with experienced AI quantitative researchers earning between $500,000 and $1,000,000 annually.
Manufacturing: Transitioning from Automation to Intelligence
AI applications in manufacturing have unique characteristics: they involve not just software-level intelligence but also the upgrade and transformation of hardware systems. This creates a demand for AI professionals who possess a hybrid skill set—knowledge of both AI algorithms and mechanical engineering, as well as data analytics and production processes.
In China, the rapid development of smart manufacturing has spawned numerous new roles: AI equipment engineers, smart manufacturing systems integrators, and industrial data analysts, all of which require interdisciplinary and practical expertise.
Education and Healthcare: Fields with the Highest Ethical Sensitivities
Education and healthcare are among the areas where ethical issues related to AI are most pronounced. In these sectors, the need for AI ethics officers is particularly acute.
The rapid evolution of healthcare AI—from imaging diagnosis and drug discovery to surgical assistance and personalized treatment—means that the demands for safety and explainability are exceptionally high, given the catastrophic consequences of medical errors.
Similarly, in education, while AI can offer personalized learning, intelligent tutoring, and automated assessments, it also raises concerns about privacy, fairness, and humanistic care.
The Salary Revolution: The "Golden Age" of AI Skills
Data indicate that AI skills are fast becoming the most valuable asset in the workplace. PwC's research reveals that the wage premium for professionals with AI skills jumped from 25% to 56% in just one year, and this trend is accelerating.
Salary Structures for AI Talent at Different Levels
- Top AI Scientists: $200,000–$500,000 per year, primarily employed by major tech companies and top research institutions, typically holding PhDs and extensive research experience.
- AI Engineers: $150,000–$300,000 per year, responsible for developing and deploying AI systems; currently the most in-demand AI role.
- AI Product Managers: $120,000–$250,000 per year, overseeing the planning and promotion of AI products, a role that requires both technological and business acumen.
- AI Collaborators: $80,000–$150,000 per year, emerging rapidly as a new role.
- AI Ethics Officers: $100,000–$200,000 per year, a growing field alongside heightened regulatory demands.
Shrinking Regional Disparities
Interestingly, the value of AI skills is transcending traditional geographic boundaries. The rise of remote work enables top AI talent to serve employers globally, narrowing salary gaps between regions.
In China, AI professionals’ salaries are rapidly catching up to those in Silicon Valley. According to Levels.fyi data, the salary range for AI engineers in China is now approximately $64,000–$134,000 (about 463,000–970,000 RMB), ranking 12th globally.
Challenges and Risks: The Shadows We Cannot Ignore
Despite the tremendous opportunities brought by AI, we must also acknowledge its challenges and risks.
Widening Employment Inequality
AI's development may exacerbate inequality in the labor market. Those who adapt and acquire new skills will secure higher incomes and better opportunities; those who fail to adapt could face unemployment or reduced earnings.
This disparity exists not only on an individual level but across regions and nations. Developed countries and regions with an advantage in AI technology and talent may further widen the gap, while developing nations could face greater challenges.
The Pressure of Constant Skill Renewal
Given the rapid pace of AI advancement, related skills must be continuously updated. This poses a significant challenge for individuals, requiring not only the acquisition of new skills but also frequent relearning.
This pressure is especially pronounced for older workers. Data show that younger people adapt more easily to AI tools, whereas those over mid-career often need more time and effort to master new skills.
Redefining Human-AI Relationships
As AI assumes an increasingly prominent role in work, the boundaries between human and machine become blurred. This is not just a technical issue but also a social and ethical one.
We need to reconsider the meaning of work, the value of human contributions, and the role of technology in society. These questions have no standard answers and require collective exploration and discussion.
Strategies for Adaptation: A Dual Approach for Individuals and Society
In facing the transformation of the workplace brought about by AI, preparation is needed on both individual and societal levels.
On an Individual Level: Proactive Adaptation and Lifelong Learning
- Reallocating Skill Investment: Devote more time and energy to skills that AI cannot easily replace, such as creative thinking, emotional intelligence, and solving complex problems.
- Deep Utilization of AI Tools: Not only learn how to use AI but also understand its underlying principles and limitations to cultivate effective human-AI collaboration.
- Cross-Disciplinary Knowledge Accumulation: In the AI era, the most valuable asset is often a combination of skills from diverse fields—capabilities that AI cannot easily replicate.
- Mindset Adjustment: View AI as a partner rather than a competitor, and learn to leverage your unique strengths in human-machine collaboration.
On a Societal Level: Policy Support and Institutional Innovation
- Reforming the Education System: Adapt educational content and methods to nurture talent suited for the AI era, encompassing not only technical skills but also critical thinking, creativity, and communication.
- Enhancing Professional Training: Provide more opportunities for adult education and skill training to help workers keep pace with technological changes.
- Improving Social Security: Establish more flexible and inclusive social safety nets to support those affected by technological disruptions.
- Establishing Regulatory Frameworks: Develop sensible AI regulatory policies that both encourage technological advancement and protect workers' rights.
A Glimpse into the Future: A New Era of Human-AI Collaboration
At the conclusion of his talk, Hassabis said, "The future of work is not about humans versus machines, but about humans and AI solving more complex problems together."
This statement points us in the right direction: the core of the AI era is not replacement, but collaboration.
In this new era, the most successful individuals and organizations will be those that master human-AI collaboration. It requires us to rethink the essence of work, redefine the roles of individuals, and redesign organizational structures.
Redefining Work
In the AI era, the value of work will increasingly lie in creativity, judgment, and emotional connection—innate human qualities. Those who can combine these strengths with AI’s computational power will hold a distinct advantage in the workforce.
Redesigning Organizations
Companies need to restructure their organizations and workflows to accommodate new modes of human-AI cooperation. This includes establishing new team structures, new performance evaluation criteria, and innovative mechanisms for collaboration.
Reevaluating Society
More broadly, the evolution of AI will prompt us to rethink how society is organized, how value is distributed, and the direction of human civilization. These are issues we must address seriously even as we embrace the conveniences AI brings.
Conclusion: Embracing Change, Shaping the Future
Hassabis’s prophecy is not merely an outlook on the future—it is a call for action in the present.
The AI revolution has already begun, and the restructuring of work is underway. Each of us faces a choice: will we passively await change, or actively embrace it?
The data show that those who master AI skills early are reaping enormous rewards. More importantly, we must grasp the deeper logic behind this transformation, find our place in the new era, and generate greater value for ourselves and society.
As Hassabis reminds us, humanity possesses "infinite adaptability."
In this unprecedented technological revolution, our ability to adapt will be tested once more. With our ingenuity, creativity, and collaborative spirit, we can not only adapt to these changes but also shape a brighter future.
The new ecosystem of work in the AI era is taking shape, and each one of us is both a participant and a shaper of this ecosystem.