The quiet crisis in many CX and EX teams today is not technology. It is confidence.
Frontline managers, contact center leaders, and HR heads feel the pressure to “become AI-first,” even as their people worry whether they still matter in a machine-led future. Employees ask if their skills will stay relevant. Leaders ask if their operating model can keep up. Boards ask where productivity gains will come from. All three questions point to one strategic imperative: workforce intelligence. EPOCH Framework is the answer.
This is where CX and EX intersect with AI in a very real way. Workforce intelligence is no longer a back-office HR exercise. It is now a frontline capability that determines how well your people, your processes, and your AI systems come together to deliver differentiated experiences.
EPOCH Framework: human capabilities still win
MIT Sloan research reframes the AI debate away from “jobs lost” versus “jobs saved” and toward “human capabilities that complement AI’s shortcomings.” The study by Isabella Loaiza and Roberto Rigobon looks at 19,000 work tasks across about 950 jobs in the U.S. labor market and introduces a framework called EPOCH Framework.
EPOCH Framework highlights five uniquely human capability clusters.
- Empathy and emotional intelligence.
- Presence, networking, and connectedness.
- Opinion, judgment, and ethics.
- Creativity and imagination.
- Hope, vision, and leadership.
When these capabilities as explained in EPOCH Framework are embedded in roles, those roles show higher employment growth and lower risk of full automation. The biggest employment lift in their data comes from “hope” and “opinion” capabilities, both central to effective CX and EX leadership.
EPOCH Framework: What this means for CX and EX roles
The same research uses risk-of-substitution and potential-for-augmentation scores to assess which tasks are likely to be automated or enhanced by AI. Tasks rich in EPOCH Framework capabilities show more resilience and are more likely to grow. That has direct implications for CX and EX design.
For CX and EX leaders, roles heavily grounded in empathy, judgment, and creative problem-solving become strategic assets, not cost centers. Contact center agents who de-escalate emotionally charged interactions, experience designers who imagine new service journeys, and people leaders who set a compelling vision for a better workplace all sit in the “more human-intensive” territory that AI struggles to replace as per EPOCH Framework.
This does not mean these roles are untouched by AI. The research notes that tasks with high automation or augmentation potential still see job loss when productivity rises and headcount stays flat. The leadership challenge is to convert those productivity gains into reinvestment in higher-value CX and EX work instead of pure cost takeout.
EPOCH Framework: Five AI issues CX/EX leaders cannot ignore
At an MIT Stone Center conference on inequality and the future of work, experts identified five workforce issues that matter as AI adoption accelerates. All five sit squarely in the CX/EX agenda.
First, leaders must understand which workers are affected, and what safeguards they have. Panelists noted that exposure to AI risk is uneven; sectors like finance and insurance are highly exposed but often have low union density, making disruption more painful for individuals. For CX, this maps directly to contact centers, back-office operations, and outsourced support partners who may not have strong protections or clear reskilling pathways.
Second, companies need to integrate worker input into AI decisions. MIT professor Erin Kelly stressed the difference between involving frontline workers in solving real problems versus asking them to rubber-stamp an already-decided rollout. For EX leaders, this means co-designing AI use cases with agents, store staff, and service teams instead of pushing down tools that feel like surveillance or thinly veiled replacement.
Fairness, transparency, and trust in AI-enabled EX
The panel also emphasized transparency and fairness, especially in hiring and promotion processes where AI is used. Frida Polli pointed out that algorithmic decisions leave an auditable trail of statistical relationships, unlike human decisions that leave no structured record. This creates an opportunity to detect and correct bias, but only if organizations are willing to measure and report on those patterns.
For EX leaders, this is a double-edged sword. AI-based hiring and promotion tools can surface hidden talent and reduce subjective bias, yet they can also encode past inequities. The experts argued that businesses should voluntarily test whether AI systems disproportionately favor certain groups and adjust models and policies where needed, instead of waiting for regulators to force their hand.
Engagement with policymakers is another critical theme. Polli shared that proactive corporate involvement helped shape New York City’s AI transparency law in hiring, balancing fairness with practical implementation. For CX-intensive industries, early engagement can ensure that regulations around monitoring, automated decisioning, and customer profiling protect workers while still enabling service innovation.
The EX risk leaders often underestimate
One of the most overlooked findings in the report is the link between AI-driven job insecurity and worker well-being. Erin Kelly reminded leaders that unemployment has been recognized as a health hazard since the 1920s, with clear connections to stress and productivity loss. Job instability and fear of replacement can erode engagement long before any formal layoff happens.
Molly Kinder highlighted a second-order risk that matters deeply for CX and EX. If AI takes over all “easy” work in professions such as therapy or medicine, humans may be left only with the most complex, draining cases, which accelerates burnout. This dynamic can surface in CX when bots handle quick resolutions and agents are left handling only irate, escalated customers all day. Without redesigning roles and support structures, leaders can unintentionally degrade EX even as customer metrics appear to improve.
How to use generative AI to augment, not erode, your workforce
MIT Sloan professor Danielle Li offers a pragmatic blueprint for using generative AI to actually augment workers. She focuses on four moves: defining success, investing in data infrastructure, incentivizing workers, and clarifying which tasks AI should substitute.
First, leaders must define what AI success looks like for their context. Large language models are “smart” in general but need clear, labeled examples of what “good” performance means in a specific company or workflow. Bank of America’s Erica, which now serves tens of millions of customers monthly, works well because its model is grounded in rich internal data and well-defined tasks, not vague AI ambition.
Second, organizations need real data infrastructure, including ways to pool data where appropriate. AI has made the most progress so far in domains like customer service, software engineering, and R&D support, where large corpora of labeled examples exist. CX leaders who fail to invest in clean interaction data, feedback, and journey outcomes will struggle to realize meaningful AI gains, no matter how advanced the tools.
Compensating and engaging your “human training data”
Li also flags a tension EX leaders cannot ignore. Employees may resist having their work used as training data if it devalues their expertise and future roles. A recorded lesson from a skilled trainer, or transcripts from a top-performing agent, can be invaluable for building AI models, yet it also raises the question: if the model mimics them, what is their long-term role?
The report argues that humans are paid because expertise is rare. When that expertise becomes abundant through AI replication, traditional pay logic comes under pressure. Li suggests companies should incentivize or compensate employees for their contributions to AI training, positioning them as active stakeholders rather than passive donors of intellectual capital.
Finally, the line between substituting tasks and substituting jobs must be drawn deliberately. Using AI to read late-night radiology images, for instance, could be safer than relying on an exhausted human, yet that does not mean removing human oversight or clinical judgment. CX leaders need similar nuance: automate repetitive scripting and documentation, but retain humans for complex negotiations, service recovery, and trust-building moments.
Why juniors cannot carry the AI change alone
Another important warning in the report concerns who leads AI learning inside organizations. MIT Sloan professor Kate Kellogg and colleagues found that relying on junior employees to teach senior professionals about generative AI is risky. Their study shows that juniors often use what the authors call “novice AI risk mitigation tactics” that differ significantly from expert guidance.
Through interviews with 78 junior consultants who used GPT‑4 on a business case, researchers identified three traps.
- Limited understanding of AI accuracy, explainability, and contextualization.
- Focus on changing human routines instead of system design.
- Focus on project-level workarounds instead of firm or ecosystem-level safeguards.
For example, juniors recommended always using AI only to polish human work, while experts focus on selecting appropriate use cases where AI error risk is acceptable and then redesigning systems accordingly. EX and CX leaders therefore need to take direct responsibility for AI governance, vendor oversight, and system design, rather than outsourcing that thinking to tech-savvy juniors.
Using AI to find and close skills gaps
If AI reshapes work, leaders need a clearer view of current skills and emerging gaps. Research from the MIT Center for Information Systems Research shows that in a 2022 executive survey, leaders estimated that 38% of their workers would need fundamental retraining or replacement within three years. At the same time, separate survey data shows employees are eager to learn: more than half said they need more training to do their job better, and over three-quarters said continuous learning would make them more likely to stay.
Nick van der Meulen’s work highlights skills inference as a promising approach. Johnson & Johnson, for example, used AI to analyze workforce data and infer proficiency levels in “future‑ready” digital skills such as master data management and robotic process automation. They created a taxonomy of 41 targeted skills grouped into 11 capabilities, then combined evidence from HR systems, recruiting, learning records, and project platforms to estimate skill levels while protecting individual privacy.
From skills intelligence to career lattices
The impact of this skills inference approach is significant for both EX and business outcomes. Johnson & Johnson saw a 20% increase in use of its learning ecosystem after the first round of skills inference. By March 2024, 90% of technologists had accessed the learning platform, using the insights to compare current skills against desired future roles. Leaders, meanwhile, could see heat maps of capabilities by region and business line, guiding where to invest in development for maximum impact.
Van der Meulen stresses that careers will increasingly look like lattices rather than ladders. Some employees will pursue traditional promotion paths, but others will opt for lateral moves, new emerging roles like prompt engineering, or even deliberate step-backs for personal reasons. To support this, organizations must go beyond e-learning libraries to offer mentorship, on-the-job practice, and visible examples of successful reskilling journeys.
A critical mindset shift is recognizing that skills are dynamic, not fixed. What makes an employee successful today will not necessarily sustain their value three, five, or ten years from now. CX and EX leaders who treat skills as a living asset, continually refreshed, will weather AI disruption more effectively than those who assume yesterday’s competencies will stretch indefinitely.
Actionable recommendations for CX and EX leaders
Translating these insights into practice requires deliberate moves across strategy, design, and governance. The following actions draw directly from the report’s findings and apply them to CX and EX contexts.
Explicitly value EPOCH capabilities in CX/EX roles
- Redesign job descriptions and performance metrics to foreground empathy, judgment, creativity, and leadership, not just handle time and volume.
- Use these capabilities as a lens when deciding which tasks to automate, which to augment, and which to invest in as signature human experiences.
Co-design AI adoption with frontline workers
- Set up structured listening forums and pilots where agents, store staff, and people managers help define AI use cases that improve job quality and customer outcomes.
- Make worker input a mandatory stage gate for any AI deployment that materially changes tasks, workflows, or evaluation criteria.
Build transparent, fair AI systems in people processes
- Audit AI-enabled hiring, promotion, and workforce analytics for disparate impact across demographic groups and adjust models or thresholds when you find skew.
- Communicate clearly what data you use, how you make decisions, and how employees can challenge or appeal AI-influenced decisions.
Invest in data infrastructure for CX-focused AI
- Clean and standardize customer interaction, feedback, and outcome data to create the labeled examples AI needs to be genuinely useful.
- Where appropriate and legally safe, explore data pooling or partnerships in non-differentiating areas to achieve the scale needed for robust models.
Compensate and recognize employees whose expertise trains AI
- Design incentive schemes, recognition programs, or differentiated career paths for top performers whose work becomes training data for AI systems.
- Make it explicit that these employees are not training their replacements; they shape how the organization works with AI and receive rewards accordingly.
Keep AI governance with senior leadership, not just juniors
- Create cross-functional AI councils with CX, EX, risk, legal, and technology leaders who define appropriate use cases, risk tolerances, and escalation routes.
- Encourage juniors to experiment and share tips, but anchor final decisions on system design, model choice, and deployment policies in expert and leadership judgment.
Use skills inference to drive a CX/EX skills agend
- Develop a future-focused skills taxonomy that covers digital, data, and human capabilities needed for next-generation CX and EX.
- Apply AI-based skills inference, with appropriate privacy safeguards, to understand current strengths, gaps, and regional or functional hotspots.
Shift from career ladders to career lattices
- Create visible pathways for lateral moves and new roles in CX analytics, journey orchestration, and AI operations, not just vertical promotions.
- Pair skills insights with mentoring and rotational programs so employees can practice new skills quickly, not just watch training content.
Protect well-being in an AI-augmented workplace
- Monitor workload intensity, emotional load, and case mix as bots take on simple tasks and humans handle complex ones, then adjust staffing, rotation, and support.
- Build psychological safety into EX so employees can speak up about burnout or ethical concerns related to AI without fear of reprisal.
Taken together, these actions move CX and EX leaders from reacting to AI toward architecting a human-centered, AI-augmented workforce. The core message from the MIT Sloan report is clear: AI will keep reshaping work, but organizations that double down on human capabilities, transparent design, and intelligent skills development will not only protect their people—they will unlock the next wave of customer and employee experience innovation.
