The Idea in 60 Seconds :

  • GenAI is accelerating one of the largest job transitions in history, with over a million Australians projected to be displaced by 2030, mostly in white-collar and administrative roles.
  • Job loss is more than economic : It disrupts identity, self-worth, and purpose, particularly among workers whose sense of value is tied to usefulness.
  • The retraining narrative skips grief: many resist not because they won’t adapt, but because they feel ashamed or invisible.
  • GenAI can perform latent skill inference from natural language, identifying real capabilities from how people talk about their past, often more accurately than resumes or self-assessments.
  • Narrative prompts can surface deep traits and hidden strengths, transforming personal stories into data-rich profiles ready for career mapping.
  • By combining narrative inference with structured job databases (like O*NET or ESCO), GenAI can map people’s existing strengths to unfamiliar but aligned roles.
  • Technology alone isn’t enough : humans are still required to interpret, deliver, and hold space for emotional reintegration during career transition.
  • With the right model-plus-human pairing, GenAI can become a mirror that helps people reclaim purpose in a changing world.

Generative AI Will Cause And Can Help With Job Losses

The world is quietly undergoing one of the largest job transitions in modern history. GenAI is displacing workers across white-collar and blue-collar roles. I wrote about some of this as far back as 2016 as well as more recently.

How Many Australians Will Be Displaced by AI—and When?

1.3 million Australian workers, approximately 9% of the workforce, are projected to be displaced by AI by 2030.

Conservative estimates indicate a net shrinkage of ~11% in Australia’s job market (about 1.5 million fewer workers by 2030), concentrated in white-collar and administrative roles.  

From Displacement to Disorientation: The Emotional Cost Of Displaced Jobs That We Sometimes Ignore

Job loss doesn’t just affect countries economically as anyone who has been fired or made redundant will tell you.

For many, work isn’t just what they do. It’s how they prove they matter. When that vanishes, they lose identity, structure, status, and often self-worth.

Traditional advice says: “Retrain. Upskill. Move on.” But displaced workers often carry silent shame. Especially men. They feel they failed. That their skills aren’t needed. That they’re becoming invisible. What looks like resistance to retraining is sometimes just heartbreak no one named. All the more when they’re replaced by a computer.

In my view, any scheme to address upskilling displaced workers would have to involve consideration of the individual’s thought patterns.

Latent Skill Recognition

The research and my own experience tells me that GenAI can identify latent skills in people faster and more comprehensively than any human ever could. Large language models have been scientifically validated to perform latent skill inference from natural language. That is, they can read a few paragraphs of someone describing their job history, weekend hobbies, and tasks they’ve enjoyed—and extract a surprisingly accurate profile of the underlying capabilities that person holds, often better than the person can articulate themselves.

Two methods have emerged from the research:

  • Zero-shot skill extraction using transformer models (e.g., BERT) on unstructured language input to infer job-relevant traits.
  • Semantic embedding models (like SBERT or OpenAI’s own models – just ChatGPT!) that match personal narratives to existing job competency frameworks with high precision.

Most people, especially those in physical or traditionally non-academic roles, don’t know how to describe their skills in the language of job markets. GenAI could help displaced workers locate and articulate those skills.

Take a fictitious, but representative, case:

Mike, 48, worked in mining for 20 years. He was recently made redundant as autonomous vehicles replaced most drivers at his site. Mike doesn’t think of himself as technical. But he ran shift logistics, worked as a diesel mechanic for a decade, and rebuilt a classic ute with his teenage son. GenAI can infer from just a few paragraphs Mike might write on these subjects that Mike understands systems of systems, real-world engineering constraints, safety-critical coordination, and root-cause failure analysis. These are exactly the kind of real-world skills needed in robotics maintenance,  and advanced manufacturing support, things the growing economy needs.

Most frameworks for career transition fail here, in my view. They rely on self-assessment, on CVs, on outdated competency checklists. But the displaced worker isn’t a blank slate. They have a history full of effort, pattern recognition, adaptation, and intelligence. They just don’t always speak in the language of HR / LinkedIN.

From Shame to Signal: What Purpose-Driven Language Reveals

Here’s what I would ask them so Generative AI  could infer their talents:

  • What parts of your job did you love?
    Their answer to this reveals core motivators. Someone who says “solving problems no one else could” indicates autonomy, troubleshooting, high cognitive demand. Someone who says “helping people get home safe” values responsibility and care.
  • What made a good day feel like a good day?
    This sort of question surfaces temperament, rhythm, reward structure. Fast-paced multitasking? Quiet craftsmanship? Mentoring others? That’s job design data which can help the Generative AI app infer important elements of someone’s job preference.
  • What hobbies or personal projects have you stuck with the longest?
    Hobbies suggest persistence, self-direction, technical depth and aesthetic preference. A 10-year side project restoring bikes shows diagnostic strength and it’s all the more important since they do it voluntarily in their spare time.
  • What do your friends and family ask for your help with?
    This surfaces unspoken competence. Mediation? Spreadsheets? Fixing broken machines? The answer to questions like this shows informal authority and skill.
  • If money didn’t matter, what kind of problems would you like to solve?
    This reveals purpose alignment. Tells you what lights them up indicating potentially where they’ll stretch and grow.

You’ll notice these questions are designed to elicit revealed, not expressed preferences – a more reliable indicator of latent skill.

The results are natural language, which, to a model like GPT, it’s as revealing as a psychometric test. GenAI can identify patterns of inference: systems thinking, attention to detail, emotional fluency, risk management, entrepreneurial orientation. It doesn’t need formal labels. It simply reads what’s there.

I would ask them to read these answers in to a verbal conversation audio snippet for ChatGPT. Their voice, tone and the cadence of their speech all reveal valuable, hidden information to the AI who is analysing them.

Give It A Go Yourself!

You can try this. Write 300–500 words answering the questions above, honestly. Then feed it into ChatGPT with this prompt:

“Analyse the following text for evidence of core skills, motivations, personality traits, and potential job fit. Be as detailed and interpretive as possible. Identify both explicit and latent signals. Assume the writer is seeking career direction after a role displacement.”

The return isn’t just job suggestions. It’s a sense of (your) self, reassembled from fragments of ordinary language.

This is what traditional systems miss: people know who they are. They just don’t know how to say it in the right way. LLMs can listen, translate and play back to displaced workers a version of themselves they might never have seen.

Mapping Insight to Opportunity: Where Human Meets Machine

Once the model understands who someone is, the next question is where they can go. It’s a mapping problem, between unspoken strengths and unfamiliar roles.

Generative AI can help substantially, here, too, but only if it has access to the right substrate: large, structured job datasets that capture real-world skill linkages. Two standouts are:

  • O*NET (Occupational Information Network) – Maintained by the U.S. Department of Labor, O*NET is a comprehensive taxonomy of job roles, broken down by required knowledge, skills, abilities, interests, and work contexts. It’s designed for workforce planning, not recruitment. So it’s rich in generalisable traits and can be cross referenced to equivalent jobs in Australia.
  • ESCO (European Skills, Competences, Qualifications and Occupations) – The European Union’s framework. Broader than O*NET in some ways, with multilingual and qualification-mapped pathways.

These datasets codify relationships between latent skills and real world job titles. For example:

  • A person with a background in logistics coordination, attention to detail, and complex scheduling (from mining site logistics) might map to supply chain analyst, operations manager, or transportation planner roles.
  • A former diesel mechanic, fluent in diagnostic thinking, hands-on systems, and risk mitigation, might map to robotics technician, mechatronics support, or autonomous systems field tester.

These mappings aren’t necessarily obvious. Most people (perhaps especially those from mining) wouldn’t see “robotics” and think, “that’s me.” But a model trained on job competency frameworks can make the leap. It knows that a diesel engine and a robotic arm both require system diagnostics, torque calculations, and troubleshooting under pressure.

When these databases are used alongside the narrative data, those two pages of honest reflection, you get career inference that’s effective because it’s precise and personalised to one.

This is the hinge point but the final step. The act of believing a new identity is possible, often needs a human. In my opinion, a human to help the displaced worker is just as important as the technology for a couple of reasons.

The Human Interface: Why Technology Alone Isn’t Enough

A PDF isn’t a solution. A dashboard isn’t a conversation. Even when GenAI produces a brilliant profile or job match, it will land flat if the person doesn’t believe it. That’s where having a human hold the displaced workers hand comes in. The difference between “This is your new future” and “Let me show you what I see in you” is important.

The most effective approach in my view would pair the model with a human face, someone kind, competent, and emotionally attuned. A trained guide who can sit across from a person and help them walk through the results, ask questions. Someone who knows when to pause. When to offer hope and when to back off.

AI can read between the lines. But only humans can help someone see the difference between a dead end and a doorway. Done well, the process doesn’t wouldn’t just place someone in a job, it could give them back a story in which they matter.

Why the Human Matters Twice

GenAI isn’t perfect. It gets things wrong. It hallucinates facts. It reflects biases from its training data.

The job of the operator is twofold:

  • Check the outputs for errors and fit.
  • Sit with the person, kindly, until they can see themselves inside the new frame.

Gender, Purpose, and How We Frame Opportunity

One-size-fits-all doesn’t work here. People bring different values to the idea of work, what it gives them, what it means, what it threatens when it’s taken away. And gender can shape those meanings.

For many men, purpose is tightly bound to their sense of their own usefulness. The loss of a job is it’s existential. A blow to identity. “What am I for, now?” When we show them a new role, it’s not enough to say “you can do this.” They need to see how it matters. How it contributes. Framing the outcome in terms of responsibility, mastery, or being of service can help it land.

For many women, the story runs differently. They often already carry multiple roles: professional, caregiver, emotional load-bearer. So the job loss might hit on other levels, independence, stability, self-worth. The reframe isn’t just “here’s what you’re good at,” but “here’s what lets you keep growing, choosing, being whole.”

The point is to deal with each person as an induvial, not a gender. And every pattern has exceptions.

In Conclusion

There will be a period of transition during which some people displaced from work will want to find another job but might be feeling disillusioned.

As well as being the cause of the displacement gen ai is uniquely positioned to help people find jobs in the new economy if we think ahead.

Any approach would benefit from a human to human interface to ensure messages are delivered only at a pace each individual can take, honed to their specific motivations and designed to highlight their strengths that the new job market values.