Evolving Roles and Responsibilities in the Age of AI
Updated: Jul 19
By Michelle L. Kissler, PMP, SHRM-CP, CCMP — Service Leader at Levata Human Performance® | June 2025
Organizations around the world are embracing generative AI agents to boost productivity, accelerate decision-making, and reduce the burden of routine work. But as AI takes over more task-based responsibilities, many leaders are left wondering: What happens to the people who used to do those tasks? And what skills are actually needed now?
The Challenge Beyond Upskilling
For years, conversations about AI in the workplace have focused on "upskilling" — a vague and often redundant term that oversimplifies the challenge. But it's time to take the next step. As entry-level roles and other task-heavy jobs transform — or vanish entirely — we must rethink what workforce development looks like in an AI-integrated organization.
Disappearing Tasks, Not Jobs
The first shift we must acknowledge is that AI is not eliminating jobs — it's eliminating tasks performed by those jobs. In many industries, the traditional entry-level role was built on executing repeatable, process-driven work: data entry, report formatting, note-taking, scheduling, and basic triage. These are the exact types of tasks AI can now perform with remarkable accuracy and speed.
This isn't new. We saw it begin during the Robotic Process Automation (RPA) wave. But RPA was costly and required upfront investment. Now, with the rise of low-code/no-code AI agents, small and mid-size businesses can adopt similar automation — at a fraction of the cost and time. What's missing? Oversight.
Who Is Checking the Work?
AI doesn't always get it right. Ask any professional who has used generative AI to create content, analyze data, or draft communications — they'll tell you: it's rarely perfect on the first try. It gets you started, but rarely fully completes the requested action with no corrections or input needed from a person.
That's why quality control, validation, and source checking are essential "new" skills. These capabilities are not always taught in traditional education or emphasized in job training programs. One might argue students should gain this experience from writing research papers or evaluating sources in college — but now, many of those papers are written with the help of AI. Rather than resisting AI use in the classroom, education systems need to shift focus: teach students to evaluate AI-generated content with a critical eye. In the AI era, these analytical and evaluative skills are as vital as technical proficiency.
Internal AI Auditors
Professionals who verify AI outputs meet quality and compliance standards.
Quality Reviewers
Specialists who test and validate AI-generated content for accuracy.
Data Integrity Leads
Experts who ensure data sources and AI results maintain reliability.
New roles are emerging that focus on this oversight. These are roles that require critical thinking, discernment, and the ability to recognize when something looks "off." It's less about knowing all the answers and more about knowing how to test the answers AI gives.
Modernizing Apprenticeships
A growing concern we're hearing from executives is about where Gen Z fits in this AI-integrated world. When task-based roles disappear and people no longer "start at the bottom," organizations are left grappling with how to build foundational knowledge and internal awareness.
"How do we teach people how the whole system works if they're no longer inside the system doing the little parts?"
We need to reimagine entry-level work. Not as a series of routine tasks, but as a hands-on apprenticeship into strategic thinking, AI oversight, and ethical decision-making.
Instead of assigning new hires to task queues, organizations could embed them in departmental pods where they learn to manage AI agents, validate results, and think critically about workflows. These roles won't look like traditional individual contributor jobs — but they might better prepare employees to become agile, thoughtful leaders.
Generation AI: Hiring the Next Workforce
It's not just that entry-level jobs with routine tasks are being eliminated — this shift is also redefining what we expect from the next generation of talent.
Some believe workers are too quick to take AI outputs at face value, especially Gen Z. And it's not hard to understand why — when we think about why we are good at critically evaluating what AI produces, it's because of our experience. We've seen how systems break down, how nuance matters, and where details get lost. Gen Z is entering the workforce without that same lived experience, which makes it even more critical to teach not just how to use AI, but how to assess its outputs.
AI literacy isn't enough — we must teach AI judgment.
How to prompt responsibly | How to trace the source of data |
How to recognize bias and hallucination | How to ask better follow-up questions |
To bridge the experience gap, organizations can reintroduce foundational learning by embedding real-time observability into AI-enabled workflows. Rather than shielding new hires from complexity, we must give them exposure to how systems behave under pressure — when things go wrong, when friction emerges, and when human discernment is essential. This means rethinking what "learning by doing" looks like in a digital environment. Where past generations learned by being in the trenches — handling data, processing exceptions, troubleshooting errors — we now need modern equivalents: environments where people witness decision points, understand tradeoffs, and build a sense of operational "normal."
Cross-Generational Gaps in AI Judgment
It's tempting to assume that AI confidence — or overconfidence — is a Gen Z problem. But the reality is broader and more nuanced.
As AI tools become more embedded in the workplace, organizations are seeing challenges arise on both ends of the adoption curve. On one end, some employees — regardless of age or experience — place too much trust in AI-generated outputs. Because the responses appear polished and well-sourced, they're often accepted at face value. This overreliance can lead to blind spots, where inaccuracies, bias, or hallucinations go unchecked.
AI Hallucinations
Confidently generated outputs that sound plausible but are factually incorrect or entirely made up.
AI Inherited Bias
Skewed outputs caused by flaws in the training data or model assumptions — often reflecting existing societal or institutional prejudices.
On the other end, some professionals hesitate to use AI at all. Their reluctance stems not from resistance to change, but from a lack of understanding and confidence in the tool. They may worry about misusing the technology, asking the wrong questions, or being perceived as behind the curve. This hesitation can leave valuable talent feeling anxious, especially when there's no structured pathway to gain fluency.
To close this gap, organizations must offer inclusive, practical training that helps employees develop the skills to spot and challenge AI flaws — not just operate the tools. That includes:
Prompting with context and specificity
Tracing the source or rationale behind AI outputs
Identifying signs of hallucination (e.g., overly generic language, broken logic, fabricated citations)
Spotting biased or exclusionary phrasing
Knowing when human review is required
Bridging the confidence gap — between over-trust and underuse — is essential for building a resilient, future-ready workforce. AI can accelerate performance, but only when guided by human judgment.
Beyond Buzzwords: Reframing "Upskilling"
What is everyone meaning when we say we need to "upskill" the workforce for the age of AI? It's about preparing people to think differently and act with agility — in roles that are still being defined.
To do this effectively, organizations need to shift from offering disconnected training modules to designing integrated learning ecosystems. These ecosystems should prioritize three key dimensions: Cognitive Agility, Relational Intelligence, and AI Fluency and Application.
Cognitive AgilityAI speeds up execution but doesn't replace human reasoning. We need to invest in training that strengthens:
These aren't just "nice-to-haves" — they're the basis of sound judgment in AI oversight, especially when no one person owns the whole system anymore. | Relational IntelligenceAs AI takes over mechanical execution, human value increasingly lies in collaboration, influence, and ethical navigation. Future-ready teams need:
Leadership development, long seen as a late-stage career perk, must now begin on day one. |
AI Fluency and Application
Basic literacy is not enough. Every role will require a working understanding of:
How AI models are trained and where they fail
The difference between generative, predictive, and rule-based automation
When and how to intervene
Even non-technical roles must learn to "speak AI" — not just to use tools, but to evaluate and guide them.
The New Skill Stack: From Compliance to Curiosity
Upskilling in this era isn't about chasing technical credentials — it's about developing a mindset of adaptive learning. The employees who thrive won't just be the ones who can code; they'll be the ones who can connect, question, and challenge what AI presents.
To support this evolution, organizations must:
Redesign onboarding programs to include AI fluency and systems orientation
Offer modular learning paths that evolve alongside tools and needs
Build cross-generational mentorships where experience meets innovation
And perhaps most importantly, they must reward curiosity — not just compliance.
We are no longer training people to do tasks. We are training them to think, to ask, to guide, and to discern. The value of the human workforce is not shrinking — it's evolving. And as AI redefines the boundaries of what's possible, it's our responsibility to redefine what it means to be prepared.
The future belongs to those who can move from task execution to tactical intelligence — from automation dependence to augmented discernment.
Let's not just adapt. Let's get better at being human.
Comments