Which Skills Will Survive the AI Onslaught?

As professors, board members, and authors focused on all things artificial intelligence (AI), anxious friends often ask, "What type of job should my child pursue?" While we refrain from offering specific career advice, we recognize the need for boards and CEOs to balance the potential for job displacement with a framework to assess which jobs will continue to be done by people, then guide talent development accordingly.

Boards and CEOs are inundated with reports projecting doomsday scenarios of job displacement due to AI. For example, according to the World Economic Forum's The Future of Jobs Report 2020, automation and AI are expected to displace 85 million jobs globally by 2025, especially in industries such as manufacturing, data entry, and customer service due to the repetitive nature and predictable processes of these fields.

AI is already disrupting many roles hitherto considered firmly within the human domain. The 2023 strike by Hollywood screenwriters represented by the Writers Guild of America underscored concerns about AI's encroachment on creative industries. Studios are increasingly using AI tools to draft scripts or assist with story development, potentially reducing the demand for human writers. While AI can mimic patterns and dialogue, the screenwriters argued that it threatens the originality and nuance that only humans can bring to storytelling. The strike raises a critical question: If even highly creative roles are vulnerable to AI, what jobs and skills will truly survive the AI onslaught?

Irreplaceable Human Dimensions 

In analyzing a wide range of articles on the topic, when a job scores high along any of the following three dimensions, it creates significant hurdles for displacement by AI:

1. Empathy. In situations that require the warmth and concern of other humans, it is best to have one help us through a challenging situation.
2. Uncertainty. AI only knows what it has experienced through training and its ability to model predictions is still rooted in the past. When someone needs to see around corners, decide on new areas to explore, or place bets, it is expected that a person will guide them, not technology.
3. Risk management. While AI can assess risks, it cannot handle every edge situation, especially when no risk can be tolerated. Examples may include situations related to life, money, ethics, and even power.

These three dimensions often come together in areas such as health care, high-risk trading, and corporate strategy. In addition, while these dimensions define a strong need for humans as agents, AI will certainly change the nature of their jobs as it adds "copilot" support.

Who is safe? 

Plotting people's roles against these three dimensions can correctly gauge where the slightest danger lies. Below are the types of roles and skills that scored highly along these dimensions and therefore have more staying power.

Emotional situations requiring empathy and human touch. Empathy is a core need in human-to-human interactions. AI can simulate certain aspects of human interaction, but it cannot fully replicate the deep emotional understanding and compassion that humans provide to one another. It is expected that humans will be at the center of the following scenarios:

• Customer service for sensitive situations. While AI-powered chatbots and virtual assistants can handle routine inquiries, customers facing complex or emotionally charged issues, such as billing disputes or service failures, often require a human representative who can listen empathetically, understand their frustration, and provide personalized solutions with a compassionate touch. For example, when a passenger was forcibly removed from a 2017 United Airlines flight, the company faced a massive public relations crisis. AI couldn't have managed the emotional fallout; human representatives were needed to handle the sensitive customer interactions with empathy and care.
• Doctor-patient relationships. In health care, patients facing serious diagnoses or undergoing surgery need more than just technical expertise. They require medical providers who can offer emotional support, listen to their fears, and reassure them. For example, during end-of-life care, doctors and nurses often provide emotional support and comfort, especially palliative care specialists.
• Emergencies. In natural disasters, accidents, or traumatic events, people look for reassurance, stability, and understanding. The human touch in emergency response—whether it's calming a panicked victim or helping families through trauma—requires empathy. For example, during Hurricane Katrina, first responders not only rescued people but also comforted distraught survivors, helping them process the trauma they experienced. Their empathetic presence, calming fears and offering emotional support, is something AI couldn't replicate.
• Long-term relationships and trust-building. Many industries, such as counseling, education, and caregiving, rely on building long-term relationships grounded in empathy and understanding. In elder care, for example, caregivers often form deep emotional bonds with patients over time, offering companionship and understanding. In cases such as dementia care, human empathy and patience are critical to building trust.

Physical work requiring hands-on judgment. Many frontline jobs require on-the-spot problem-solving, where the worker assesses unique and variable situations in real time and translates that judgment into physical action. While AI can provide diagnostic tools or assist in finding the source of a problem, most people would prefer the reassurance that comes from having a trained professional who can gauge trade-offs and address unforeseen complications. The following are some examples:

• Plumbers dealing with an unexpected leak need to adapt to the current conditions, navigate irregular spaces, assess aging infrastructure, and adjust solutions on the fly. This requires technical skill and human judgment.
• Electricians working in diverse environments must assess unique systems, troubleshoot malfunctions, and ensure safety for themselves and their customers. Each job presents new challenges, such as interpreting wiring issues or solving irregular installation problems.
• Construction site foremen constantly make decisions based on site conditions, the weather, safety regulations, and team coordination. While AI can help with planning and logistics, a foreman must handle the daily, on-the-ground complexities, such as adjusting plans for unexpected obstacles or ensuring safety protocols in real time.

Jobs that require envisioning the future. While AI can analyze data and generate options, humans are still essential to seeing around corners and looking ahead to ensure that strategic goals, values, and long-term ambitions are aligned. This is not just an executive-level skill but one that is critical for any employee who deals with ambiguity, uncertainty, and tough trade-offs. Boards will should push management to emphasize talent development in areas of heightened importance, such as the following:

• Strategic thinking that can adapt to dynamic environments. In volatile, uncertain markets, human vision is vital to continuously reshaping and adapting business strategy. During the 2008 financial crisis, human leadership was critical to responding to rapidly changing market conditions. While AI-driven algorithms failed to foresee the collapse, human decision-makers, including central bankers and corporate leaders, adjusted policies, secured liquidity, and restructured businesses to navigate the economic turmoil.
• Navigating scenarios with no data or precedent. When tackling entirely new challenges or entering uncharted markets, human vision is critical because AI relies heavily on historical data. During the COVID-19 pandemic, governments, businesses, and health-care systems had to navigate an unprecedented global crisis without prior data on how to do so. Human leaders had to rapidly assess, innovate, and make critical decisions, including implementing lockdowns, pivoting business models, and rolling out vaccines, based on limited, evolving information.
• Design thinking that shapes new product development and end-to-end customer experiences. Creating a seamless, engaging, and emotionally resonant customer journey requires human vision to understand the complexities of customer needs and desires. A recent study by Upwork noted that while demand for basic, very low-priced creative workers had dropped, most of those workers were now moving up to doing higher-level design assignments and charging more for their services.

Decisions that have a zero margin of error. In high-stakes environments in which even the slightest error can lead to catastrophic consequences, there is no room for risk. While AI and automation have made incredible strides in reducing errors and improving efficiency, edge cases, or rare, high-pressure scenarios, demand human judgment and intervention to ensure safety and strategic decision-making.  Humans will likely continue to manage the following scenarios:

• Autopilot systems. Automation is highly advanced in aviation: the first autopilot was introduced by Sperry Corp. in 1914, and by the 1930s, commercial airliners began incorporating basic autopilot systems. However, all commercial planes have two pilots to manage the rare, high-stakes scenarios for which the cost of failure is too high, such as sudden system failures, extreme weather, or emergencies. Safety is paramount and even a tiny margin of error is unacceptable.
• Nuclear power plant operations. Nuclear plants rely heavily on automation for monitoring and managing daily operations, but human operators are crucial to handling rare, high-stakes situations, such as system malfunctions or radiation leaks. In 1979, the Three Mile Island nuclear disaster in Pennsylvania highlighted the importance of human operators in crises and of providing them with clear, consistent user interfaces to see what's going on. Although the plant relied on automated systems, human intervention was critical when a cooling malfunction caused a partial reactor meltdown. Due to the hodgepodge of suppliers and proprietary gauges in the control room, operators had to rapidly triage the issues and manually stabilize the reactor to prevent a catastrophic radiation leak, demonstrating the irreplaceable role of human judgment in high-stakes scenarios.
• Financial trading during market crises. High-frequency trading algorithms dominate much of the financial market, executing transactions at lightning speed. However, during periods of extreme market volatility, such as financial crashes or geopolitical crises, human traders step in to manage risk and make strategic decisions. During the 2010 flash crash, high-frequency trading algorithms caused extreme market volatility, wiping out nearly $1 trillion in stock value within minutes. Human traders had to step in to stabilize the market, halting algorithmic trading and restoring order.

Roles that strengthen existing power structures. Historically, power tries to reinforce itself, adapting to new technologies rather than being disrupted by them. AI will follow a similar path, where those at the top of existing power structures (e.g., corporate leaders, wealthy investors, and government officials) will use AI to maintain and expand their influence, ensuring that disruption strengthens their control rather than undermines it. Humans will explicitly control AI in several areas where its impact determines the balance of power:

• Government as the ultimate authority. Governments, which hold the ultimate power over regulation and policy, will continue to run nations, with AI acting as a tool under their control. AI may enhance the efficiency of government services or national defense. However, the decision-making and authority will remain firmly in the hands of human leadership, ensuring that political structures persist alongside technological change.
• Control of AI means control of capital. Since AI is a form of capital, those who already control large pools of capital, such as private equity and venture capital funds, will dominate by investing heavily in AI start-ups and innovations. These entities will invest in, own, and direct AI innovation, maintaining their influence over industries as they evolve.
• Influence over societal rules and frameworks. Those at the top of existing power structures will also shape the ethical, legal, and societal frameworks for AI use. As they define the rules of engagement for AI technologies, such as privacy laws, labor regulations, and ownership rights, they ensure that AI serves their interests, reinforcing their dominance within the global power hierarchy.

Anticipate the Transition to a Human-AI Hybrid Workforce 

As AI takes over more routine tasks, companies will become leaner. Some layers in the organizational pyramid will consolidate or shrink, but the remaining roles will demand deeper judgment and higher levels of adaptability. This need for human oversight, leadership, and judgment will ensure the pyramid doesn't disappear; it just gets narrower at each level.

For boards and CEOs, now is the time to take a hard look at workforce evolution by doing the following:

• Score major clusters of jobs by how they deliver empathy, help manage risk, or require envisioning the future. Assess the vulnerability of key employee roles and run scenarios for the degree of retraining, job redesign, or job displacement that the organization is willing to pursue.
• Invest in human capital to complement AI. Prioritize identifying the talent and key places to upskill the workforce for areas where AI cannot compete, including in the roles that score high on visionary leadership, empathy, and judgment. Organizations that focus on cultivating these human-centric skills will be in charge, positioning themselves to thrive in a world in which AI dominates routine tasks.
• Assume that everyone will have AI by their side and plan accordingly for a significant organizational transition. The future workforce will blend humans and AI, each playing to their strengths. When looking at the skills described above, the three dimensions generally favor people who can take a broader view, combine emotion with effectiveness, bring multiple points of view to bear, and see the downstream ramifications of a decision. Roles requiring specialization to operate and execute are extremely vulnerable, while those requiring more problem-solving and empathy are less so. In many functions, specialization has gone quite far, such as in marketing, finance, human resources, and operations. With an AI copilot, many of the specialized operational aspects of a job are displaced by AI, so employees will need to assume the role of decision-makers and problem-solvers. CEOs should redesign their organizational structures to assume more collaboration between human judgment and AI efficiency, challenging functional areas to build more generalist, strategic thinking skills and ensuring that the human element continues to drive innovation and strategy.
• Champion ethical AI and governance. As leaders, CEOs and boards should take responsibility not only for adopting AI but for doing so ethically. While boards oversee privacy, the security of data, and fair outcomes, their role now also includes shaping AI governance frameworks to ensure that the power structures they control align AI with broader societal goals, safeguarding trust and accountability in a technology-driven world. CEOs and boards should articulate a clear, compelling vision of how AI will enhance, not replace, the human potential within their organizations and grow the range of opportunities ahead.

While no one can predict which jobs will be safe from AI, boards and CEOs have a crucial role in guiding their organizations through this transformation and fostering the skills that will withstand automation. To help guide your company through the digital future, the time to have this conversation is now.