← All posts

My take on the future of work

· Thoughts

Welcome to Pondering Careers – if you’re new, then welcome! This is my space for sharing (and organising) my thoughts on topics related to careers and the world of work. If you’ve been here before then welcome back and thank you for continuing to share this space with me.

This week I’m coming off the back of a HECTIC month; you’re probably sick of hearing about it by now, but my eldest has just finished school and it has been one heck of a ride. The good news is that it’s all over, and as I’m writing this at 12pm he’s probably still snoring in bed, thoroughly enjoying his well-deserved break. Which means that I, too, have more head space to go back to actually enjoying Pondering Careers, and as a reward for making it through I’m going to allow myself to write about something I personally find fascinating – the future of work.

This is just my take, and despite my best efforts I am yet to secure a crystal ball for myself, which means that this may all be wildly inaccurate. Take it (and everything else you read about the future of work) with a gigantic grain of salt – this is a place to hypothesise about what might be, and reflect on what has already happened, not to make bold declarations about what will or won’t happen.

So today I’m going to start out by putting things into a bit of perspective, then we’ll look at what we can learn from the past (i.e. we’ve been here before), I’ll go over ten of the emerging industries I think careers people should be aware of, and at the end I’ll share three tips for things you can do to stay in the loop.

Side note: get a second opinion

There are lots of awesome people talking about this as well, so I would recommend getting an alternative perspective to mine as well; Chris Webb has been sharing some interesting research about who’s voices show up in LLMs like ChatGPT, and who is excluded.

Chris is probably one of the first people I noticed talking about AI and its impact on our profession, and along with Danny Mirza FRSA, MSc, MA, MBA has been leading the way with experimenting and pushing the boundaries of what some of the new technology is capable of. Associate Professor Deirdre Hughes OBE is also doing really innovative work in this space with CiCi, the careers chatbot which has been designed with safeguarding measures in place.

Let’s get some perspective

Before we dive into anything else, we need to avoid conflating AI with unemployment as the primary driver of workforce change.

It’s no surprise that AI as a job-thief is fantastic clickbait – there are a plethora of articles out there claiming that AI is going to disrupt the workforce and take all of our jobs, but the evidence for these claims doesn’t really stack up.

Take this article from the ABC; they interviewed a hairdresser who wants a higher paying, less physically demanding job in sales or customer service, and who is struggling to find work. The causes of her difficulties are much more likely to be the recent rise in unemployment (meaning more people looking for work) than AI taking all the jobs, yet AI is prominent in the headline.

Job mobility is really low right now – people are struggling to switch roles, employers can’t find staff with the right skills, and the whole system feels stuck. AI is playing a part in this, but it’s not the only thing going on, and honestly, a lot of the research shows it’s actually creating more jobs than it’s eliminating. These tend to be pretty good jobs too – ones that require critical thinking, varied skills, and the kind of work humans actually enjoy doing.

There are three key mechanisms through which AI is reorganising the workforce: augmentation, hybridisation, and specialisation.

Augmentation vs. Automation

There’s an important distinction we need to make here: augmentation versus automation.

Automation is where technology takes over an entire task with very little human involvement – think self-checkouts at the supermarket or automated warehouse systems.

Augmentation helps a human do their work better or faster – like real estate agents using AI to write compelling property descriptions, or doctors using AI to analyse medical imaging more quickly.

According to Jobs and Skills Australia’s recent report on generative AI, augmentation is actually much more common than automation across most industries. When you look at the exposure different occupations have to AI, you’ll see that for almost every field, the augmentation potential (shown in purple in the chart below) is significantly higher than the automation potential (shown in green).

What does this mean for students? Rather than AI taking their jobs, they’re much more likely to be working alongside AI tools that make their work faster, more efficient, or more creative. Those augmented roles still need humans – humans with strong communication skills, leadership abilities, and critical thinking. Social services, healthcare, education, and community work aren’t going to be easily replaced, because these fields fundamentally rely on human connection and judgement.

Hybridisation

As AI automates discrete tasks which are common to multiple jobs, we expect some jobs to merge, and this is called hybridisation. The tasks that remain make the jobs more alike, hence, they merge multiple roles into one. We’re already seeing hybridisation take place across industries, and this is generally what happens with any kind of technological transformation (not just artificial intelligence).

Think about the skills that an agricultural manager has now, compared to the number of people who would have been required to handle all those different tasks in the past. A modern farmer is often part producer, part data analyst, part mechanic, part logistics coordinator, and part safety and compliance lead. The job has not become smaller, it has become broader, with technology taking over the repetitive monitoring and coordination tasks, and leaving one person to make higher-quality decisions faster.

In practical terms, a single farm manager can now handle work that previously relied on a larger team, including:

  • Monitoring livestock location and movement using GPS tags, virtual fencing and remote dashboards, instead of multiple people riding out to physically check mobs
  • Tracking animal health risks through weight, activity, temperature or behaviour signals (and automated alerts), rather than relying only on periodic visual inspection
  • Checking water levels and flow via tank and trough sensors, remote pumps, and automated shut-off or fault alerts, instead of daily manual checks across properties
  • Managing pasture and crop conditions using satellite imagery, drones, and mapping tools to spot feed availability, pests, or stress early
  • Forecasting and responding to weather through hyperlocal weather stations, predictive models, and automated notifications to plan work windows and protect stock
  • Automating irrigation and fertiliser decisions with soil moisture probes, variable-rate application maps, and scheduling tools
  • Coordinating machinery maintenance using telematics, service alerts, and predictive maintenance data (rather than waiting for breakdowns in the paddock)
  • Running biosecurity and compliance documentation with digital records, traceability systems, and audit-ready logs that reduce paperwork burden
  • Managing supply chain and logistics (ordering feed, booking transport, scheduling deliveries, tracking inventory) through integrated platforms rather than phone calls and manual ledgers
  • Overseeing workplace health and safety with incident reporting apps, lone-worker check-ins, geofencing alerts, and digital inductions
  • Managing finances and business strategy using farm management software for budgeting, scenario planning, and performance reporting across paddocks, herds, or seasons
  • Supervising contractors and seasonal staff with digital job allocation, route planning, and task verification (photos, timestamps, sensor data)
  • Engaging with advisers and markets by sharing real-time data with agronomists, vets, banks, and buyers to make faster, evidence-based decisions

In other words, work that was previously handled by a team is now easily managed by just one or two people.

New Specialisations

At the same time, there’s evidence that AI is actually creating entire new categories of jobs that combine skills in new ways, beyond what we’ve seen before.

As organisations use AI and other digital systems to produce, personalise, distribute, and measure content at scale, the work does not simply become more technical. Instead, specialist tasks create demand for new combinations of skills and knowledge, blending creative judgement, audience insight, ethics, and platform literacy alongside digital capability. In other words, we are seeing new specialisations emerge because the problems we are solving are new.

The exciting thing about this mechanism of change is that often these new specialisations are in fields where lots of people want to work – in creative roles, where we get to be innovative and strategic. Here are just a few examples of new jobs that fit into this category:

  1. Online Community Managers – As brands develop strong online communities, they are looking for people who can manage the tech platforms and the data while also speaking authentically to customers and creating engaging content.
  2. Accessibility Specialist – Accessibility is no longer a ‘nice-to-have’ for online platforms; which means there is demand for people who can combine strong technological skills with an understanding of user-centred accessibility frameworks. These people ensure that websites comply with guidelines and compliance frameworks, and can fix software when there are issues.

We’ve actually been here before

If there’s one thing that will help calm the panic about AI taking all our jobs, it’s understanding that we’ve heard all this before. Multiple times. And we’re still here, still working, still inventing new jobs we couldn’t have imagined.

AI is getting all the glory right now, but it’s actually not all that new, and it’s just one part of emerging tech. Kids need to know about all of it.

The 1950s-60s: Extraordinary Optimism

AI as a formal field was born at the Dartmouth Conference in 1956, when researchers including John McCarthy and Marvin Minsky gathered for what they optimistically called a “two-month, ten-man study.” They genuinely believed a small group working for just eight weeks could crack the fundamentals of artificial intelligence.

The predictions that followed were remarkably confident:

  • 1958: Herbert Simon and Allen Newell predicted that within ten years a computer would be world chess champion. (Reality: It took until 1997 for Deep Blue to beat Kasparov.)
  • 1965: Herbert Simon declared that “machines will be capable, within twenty years, of doing any work that a man can do.”

At the 50th anniversary celebration of that Dartmouth conference, John McCarthy himself admitted: “AI was harder than we thought.”

The 1970s-80s: The First AI Winter

When these predictions failed to materialise, funding dried up dramatically. The 1973 Lighthill Report in the UK was particularly damning, leading to the complete dismantling of most AI research. DARPA also significantly reduced US funding. The term “AI Winter” perfectly captures what happened: enthusiasm froze over, research ground to a halt, and the field became something of an embarrassment.

Then came Japan’s grand plan. The Fifth Generation Computer Systems Project, launched in 1982 with approximately $850 million, aimed to create computers that could carry on conversations, translate languages, and reason like humans. This caused panic in the West – the Reagan administration launched a billion-dollar Strategic Computing Initiative to counter Japan’s lead.

Within a decade, another AI Winter had set in, and the Fifth Generation Project ended not with a successful roar, but with a whimper.

The 1990s-2000s: The Long Quiet

The second AI winter lasted from the late 1980s through to around 2000. Corporate AI divisions shut down, specialised hardware became obsolete, and expert systems proved brittle and difficult to scale. Work continued quietly on machine learning and neural networks – but without the hype, funding, or grand predictions.

The 2010s-Present: Deep learning changes the game

What’s different this time? Data and computing power. The combination of massive datasets, powerful GPUs, and techniques like deep learning have produced genuinely impressive results. ChatGPT and other large language models can do things that seemed impossible even five years ago.

But we’re still seeing the same pattern of hype and overconfidence. The difference is that this time, the tools are accessible to ordinary people, which makes the impact feel more immediate and threatening.

The lesson from all this is that progress is slower, messier, and less predictable than experts think. Every prediction of “human-level AI in X years” has been wrong so far. That doesn’t mean AI won’t continue to advance – it will – but the timeline and impact are much harder to predict than the headlines suggest.

For a deeper dive into this history, the AI Index Report from Stanford tracks both historical trends and current developments.

Emerging tech you need to know about

AI is getting all the glory, but it’s actually just one part of a much bigger story about emerging technology. AI spans most of these emerging fields, so we need to treat Generative AI as a separate field in itself, but also recognise how it intersects with everything else.

Some quick definitions before we start:

  • Artificial Intelligence is a broad concept of machines mimicking human intelligence
  • Machine Learning is a type of AI that uses algorithms to learn and adapt to the data
  • Generative (or ‘Gen’) AI is a type of AI that generates new content

Generative AI

Let’s start here, since it’s what everyone’s talking about. Generative AI (or Gen AI) refers to AI systems that can create new content – text, images, code, music, video. Think ChatGPT, DALL-E, Midjourney, GitHub Copilot.

This technology is already reshaping how people work. Marketing teams use it to draft campaigns, programmers use it to debug code, teachers use it to create lesson plans. The key thing to understand about Gen AI is that it’s brilliant at producing plausible-sounding content quickly, but it still needs human oversight, creativity, and critical thinking – or it risks becoming AI slop (which is incidentally Macquarie Dictionary’s word of the year). It can augment human work, but it can’t replace the judgement, ethics, aesthetic sensibilities, and contextual understanding that humans bring.

Robotics & Automation

This industry is all about physical machines (robots) that can perform tasks autonomously. This shows up in manufacturing (robotic assembly lines), logistics (Amazon warehouse robots), agriculture (autonomous tractors and harvesters), and increasingly in our homes (robot vacuums, lawn mowers).

The shift here is from manual labour to machine development and maintenance. Someone still needs to design, program, install, and supervise these systems. Modern robots often use computer vision and machine learning to navigate and make decisions – which is where AI intersects with robotics.

Internet of Things (IoT)

The Internet of Things is a network of interconnected devices that collect and share data. This includes wearables (fitness trackers, smartwatches), smart home devices (fridges, thermostats, security systems), connected vehicles, healthcare monitoring, and agricultural sensors.

All this data needs to go somewhere, and that’s where AI comes in – using machine learning to make sense of the patterns and make predictions. We need technicians to develop, install, and maintain these sensor networks, and we need analysts to interpret and translate the data into actionable insights.

Digital Twins & Simulations

A digital twin is a virtual replica of a physical object or system that can be fed real-time data to run simulations for analysis and prediction. This technology is used for monitoring weather patterns, predicting consumer behaviour, planning tactical approaches in military or emergency services, and even selecting personalised medical treatments.

Think about it: doctors could test different treatment options on your digital twin before deciding which one to use on you. Engineers can stress-test a bridge design thousands of times virtually before building the real thing. AI helps create and analyse these simulations at scale.

Extended Reality (XR)

Virtual, augmented, and mixed reality tools that overlay or replace the environment. This shows up in medical training (surgeons practising procedures in VR), architecture (clients walking through buildings before they’re built), entertainment (VR gaming), and yes, career education (students experiencing different workplaces virtually – thanks to companies like Work Window).

Creating XR experiences requires a combination of creative and digital skills along with domain-specific knowledge. An architect needs to understand both architecture and the XR tools. A medical trainer needs to know both anatomy and how to design effective VR experiences.

Advanced Manufacturing

Using high-tech methods and materials to enhance productivity and performance. This includes 3D printing for aerospace components, custom prosthetics, construction materials, and even food production.

We’ll need more technicians, process engineers, and designers who can translate digital solutions into physical production. AI plays a role in optimising these processes and identifying defects.

Biotechnology

Using biology and genetics to create personalised medical and health products. This shows up in vaccine development, pest-resistant crops, gene therapies, and biofuels.

There’s huge demand for lab and clinical workers who can combine data analysis, scientific knowledge, and ethical reasoning. AI helps analyse genetic sequences and predict how modifications will work, but humans still need to interpret the results and make ethical decisions.

Clean Energy & Storage

This includes integrating renewable power into the grid, storing energy efficiently, and building the infrastructure for electric vehicles. We’re talking about energy generation and transmission, EV charging networks, and grid upgrades.

There’s massive demand for trades (electricians, engineers), planners, environmental specialists, and project managers to handle the energy transition. AI helps optimise energy distribution and predict demand, but we need humans to build and maintain the physical infrastructure.

Cybersecurity

Systems and tools to protect data, online systems, and personal information. Required in healthcare, education, banking and finance, and government – basically everywhere.

This requires more than just IT skills. There’s growing demand for compliance specialists, risk analysts, and people who can communicate security issues to non-technical stakeholders. AI is used both by cybersecurity professionals (to detect threats) and by bad actors (to create more sophisticated attacks).

Web3 & Decentralised Systems

Web3 is a new iteration of the web that’s more decentralised – people own and control their own data, identity, and assets using blockchain technology instead of relying on big platforms.

This shows up in microcredentials (verifiable qualifications stored on blockchain), cryptocurrency, property transfers, and even fine art authentication. It requires skills in ethics, compliance, data networks, and legal frameworks. AI can help verify transactions and detect fraud in these systems.

What does all this mean?

All of these technologies are developing simultaneously, and they’re all interconnected. A student who understands the basics of AI, data analysis, and one or two of these fields will have opportunities across multiple industries. The jobs aren’t disappearing – they’re evolving and multiplying.

Predicting the future of work

Now we need to talk about predictions, but I want to do this carefully, because as we’ve just seen from our history lesson, AI researchers have a pretty terrible track record when it comes to predicting the future.

The Futures Cone

There’s a framework called the futures cone that helps us think about predictions responsibly. Developed by futurists, it acknowledges that the further forward we look in time, the less sensible it becomes to make big, confident predictions.

The cone starts narrow at “now” and expands outward through different types of futures:

  • Probable futures – what’s likely to happen based on current trends
  • Plausible futures – what could happen if certain conditions change
  • Possible futures – what might happen if we really stretch our imagination
  • Preferred futures – what we want to happen

The further out we go, the wider the cone gets, because more variables come into play and our predictions become increasingly uncertain.

So rather than telling you ‘this is definitely going to happen’, I’m going to share what I think are probable developments at different time horizons, with the caveat that I’m likely to be wrong about the details.

Looking ahead to 2027

In the next two years, we expect LLMs will become integrated into everyday life and study, much like calculators or search engines did before them (remember google is now a verb). We’ll have stronger rules and regulations around Gen AI use, particularly in education and professional contexts.

AI tools will handle more routine administrative tasks – scheduling, basic data entry, initial drafts – while quality control and human oversight become recognised professional skills in their own right. I think we’ll also see growing interest and enrolment in courses focused on augmentation jobs, where people learn to combine AI tools with human expertise.

Five years out to 2030

Looking five years ahead, I expect we’ll see less focus on general-purpose LLMs and more on specialised AI agents that handle discrete, specific tasks – a research agent, a coding agent, a design agent. We’ll probably see more hybridisation of jobs, where roles that previously existed separately start to merge as AI handles the overlapping administrative tasks.

Employers will increasingly want to see evidence of actual skills and portfolios rather than just certificates. I often get asked about the impact of AI on recruitment processes, and while we really can’t predict what the landscape will look like in 5 or 10 years time we can look at what’s happening right now. Someone who can show evidence of having completed relevant projects, or been employed in a similar role previously, is a much safer pick than someone who can only provide a certificate or degree, especially when the employer doesn’t know who has used AI to write their application and who hasn’t. I think it’s safe to assume that people who can provide concrete evidence of their skills and experience will have an advantage in that landscape.

A decade from now in 2035

Ten years out, stackable, modular learning will probably become the norm. Rather than getting one degree and being done, people will regularly add new micro-credentials and specialisations throughout their working lives, from vocational and university pathways.

Demand will increase for critical thinking and creativity as AI handles more routine cognitive work but still struggles to do the more complex critical thinking. Connected systems will have expanded dramatically – IoT, digital twins, integrated cyber systems, Web3 applications will all be far more prevalent, and I wonder if school-based internships will become available even for ATAR pathways, as the line between “academic” and “vocational” education continues to blur.

Looking twenty years ahead to 2045

This is where the futures cone gets really wide, so these are more like general ideas rather than predictions. The evidence suggests that the trend is away from fixed, regular full-time employment and towards gig and contract work, and I think that by 2045 it’s more likely that we will be working for ourselves, even as a sole trader or contractor, than that we will be working for someone else.

If I’m right, and this is the case, then key transition points go out the window, replaced with continuous and gradual transitions as we constantly prepare for the next move. We will all need to become much better at showcasing our skills and proactively seeking out new opportunities, rather than hoping they will land on our doorstep.

I could be way off

These are educated guesses, not prophecies. Keep in mind that the 1956 Dartmouth researchers thought they’d ‘solve’ AI in a summer, and we’re still here almost 70 years later trying to get it to produce a photo of a person without six fingers.

As an experiment, I wanted to look back at what OpenAI (they run ChatGPT) had themselves predicted – I found this article from early 2019 (almost 7 years ago), and they were disturbingly accurate:

Large, general language models could have significant societal impacts, and also have many near-term applications. We can anticipate how systems like GPT‑2 could be used to create:

  • AI writing assistants
  • More capable dialogue agents
  • Unsupervised translation between languages
  • Better speech recognition systems

We can also imagine the application of these models for malicious purposes⁠, including the following (or other applications we can’t yet anticipate):

  • Generate misleading news articles
  • Impersonate others online
  • Automate the production of abusive or faked content to post on social media
  • Automate the production of spam/phishing content

Literally everything in this list has come true.

So, since I have no alternative crystal ball, I asked ChatGPT to predict what it might be capable of in another seven years, and this was the response:

My best predictions for “another 7 years” (so, roughly 2032)

Below are forecasts phrased as likely directions, not promises. I’m focusing on what would be observable in practice.

  • More natural multi-format interaction (text, voice, images, and video working together).
  • A shift from “chatting” to “doing”, with models completing bounded tasks via tools and workflows, not just drafting text.
  • Better trust features: clearer sourcing, more transparent reasoning trails, and fewer confident mistakes.
  • Stronger safety and governance: tighter guardrails, monitoring, and staged or constrained release for higher-risk capabilities.
  • More explicit, permission-based personalisation, with clearer user control over what is stored or used.
  • Still unlikely: perfect factual accuracy in open-ended conversation.
  • Still unlikely: a single model cleanly replacing whole jobs end-to-end in complex, human-heavy environments.

Spooky.

What do we need to do?

Right, so if we can’t predict the future with confidence, what can we actually do to help students prepare? Three things: stay calm, stay informed, and think local.

Stay Calm

First and foremost, we need to stop catastrophising. When students come to you anxious about AI taking all the jobs, avoid the hype and be positive with them. It’s ok to be aware that AI poses a real and significant risk, while at the same time not pushing young people towards feeling helpless. I wrote more about this back in Edition 69 if you want to know more, or check out the work that Alex Zarifeh is doing with his Futures Readiness project.

Instead, focus on these key messages:

  1. We expect task replacement, not job replacement. Most jobs involve multiple tasks. AI might handle some of these (the repetitive, routine ones), but humans are still needed for judgement, creativity, relationships, and complex problem-solving.
  2. Encourage students to build experience and skills. The best preparation for an uncertain future is to be adaptable, curious, and capable. Students should be trying new things, learning how to learn, and developing both technical and human skills.
  3. Share examples of job creation. Throughout history, new technology has created more jobs than it destroyed. The jobs might look different, but they exist. Share examples of new roles that didn’t exist 10 years ago: social media managers, user experience designers, data scientists, drone operators, renewable energy technicians.
  4. Emphasise augmentation. Help students see AI as a tool that will make them more effective, not a replacement for them. The people who thrive will be those who learn to work with AI, not those who try to compete against it or ignore it entirely.

Stay Informed

The landscape is changing quickly, so you need to actively keep up. Here’s how:

Set up news alerts for topics like “artificial intelligence employment,” “future of work,” “emerging technology,” and “AI policy.” This will help you catch major developments without having to actively search for them.

Read the reports (or at least scan the summaries). There are some excellent reports that come out regularly:

Critique what you’re reading. Not everything you read will be accurate or unbiased. Ask yourself: Who funded this research? What’s their agenda? Are they selling something? Is this based on actual evidence or speculation? Are they confusing correlation with causation?

Look for good news stories. Announcements about new technology, start-ups, university collaborations, and research projects. Share these positive examples with students.

Think Local

Finally, and perhaps most importantly, think about what’s happening in your own area.

Track local tech-related projects and companies. What’s being built in your region? What industries are growing? Which companies are innovating? These are the places that will hire your students, so you need to know what’s happening. Follow local business news, attend Chamber of Commerce events, and connect with economic development organisations.

Attend innovation events when you can. Things like innovation festivals, start-up showcases, and industry conferences give you a chance to see emerging technology in action and talk to the people developing it. These conversations will give you insight you can’t get from reading reports.

Be aware of new local courses and degrees. Universities and TAFEs are constantly developing new programmes in response to industry demand. If your local institution launches a new degree in cybersecurity, or robotics, or data science, that’s a signal about where the opportunities are.

Ready for the future?

To be honest, I don’t know exactly what the future holds. Nobody does.

If I was able to tell you exactly which skills students will need in the future then I would be a very popular person, but the reality is that none of us can say with any kind of certainty.

What we do know is that technology (of any type, not just AI) changes how we work, but it doesn’t eliminate the need for work. Looking at the past, we can guess that change will take much longer than we expect, and impact far fewer of us than the news media would have you believe right now.

Our role as Careers Practitioners and Educators is not to predict exactly what jobs will be in demand, or even steer students in one direction or another. We are here to help young people find and evaluate options, then make decisions about their next steps, all while building the skills they’ll need to manage their career choices long into the future. If we can, we also help them build broad, transferable skills they can use in any workplace, and boost their confidence in their strengths so they know what they are capable of.

In order to do this, we need to stay informed and engaged with emerging technology and industries, so that we can have powerful career conversations whenever they arise.

The future of work isn’t something that happens to us. It’s something we all create together. And that’s actually quite exciting.

← All posts