Redesigning Work: Are Business Leaders Asking the Wrong Question?
Rob Slane
Elena Magrini
Head of Global Research

If you had a dollar for every time you’ve read a piece with the words “AI is changing the world of work,” or some variation, you’d probably be sorted for next summer’s vacation. But even though there are hundreds of articles out there saying this, knowing that things are changing doesn’t help you get to the nub of the questions that inevitably arise from it:
A) How can my organization understand these changes?
B) How can we prioritize our response?
Answering these questions is complex. Most of the conversation around AI’s impact on work treats what can happen to jobs as a binary: they can decline in demand or grow. But as we'll show below, the content of the job itself can change – it can broaden or narrow – and both the demand and the scope of work can change at the same time, sometimes even in opposite directions.
What is a Job?
To understand how jobs are changing, we first need to answer an even more basic question: what is a job?
It’s not uncommon to hear people speaking of jobs as “a bundle of skills” or “a collection of tasks.” But these sorts of definitions should leave us cold, because they contain a glaring omission: people.
At its heart, a job is a collection of tasks requiring a particular set of skills, performed by a person. People work. They do so by performing tasks. And to do those tasks they use skills.
The Wrong Question
Why is this definition so important? Because in the age of AI, we find that while technology can increasingly perform some tasks associated with an occupation, that doesn’t necessarily mean it can or even should replace the person doing the job.
Yet, for many businesses, that’s precisely where the conversation about AI has been. Based on the assumption that “AI could do the job”, many companies laid off workers. Yet, as the research and consulting firm Forrester found, more than half of employers they surveyed regretted AI-related layoffs, while a survey by Orgvue found that among organizations that eliminated roles following AI implementation, 32% ended up rehiring staff.
Why did they regret the decision, and why did almost a third end up rehiring? The most likely explanation is they found that AI rarely – if ever – replaces whole jobs. What it does is automate or augment some tasks within certain jobs.
Companies that tried to replace people with AI were fundamentally asking the wrong question:
“Which jobs can we axe and get AI to do?”
Whereas the real question that needs to be asked is:
“Which tasks can AI help us perform more efficiently, so that our people can spend more of their time on the tasks that create real value?”
The first question is appealing, because it is so simple. All you need to do is see a job that looks like it could be automated by AI and cut it. Yet the simplicity of the question masks the far more complex truth, which is that AI does not generally replace whole jobs, but automates or augments particular tasks within them.
The second question is far harder to answer, because it involves unpacking jobs, sorting them into their constituent tasks and skills, and exploring which of these can be helped by AI and which cannot. Who has the time or the know-how to do all that?
Understanding Tasks to Redesign Work
The starting point is understanding what your people actually do. Which tasks make up their jobs? How much time do they spend on each? Which of these could potentially be automated, augmented by AI, or continue to require human expertise?
This is precisely what our Work Intelligence enables organizations to understand. By connecting task-level intelligence with Lightcast's jobs and skills taxonomies and evidence from the labor market, organizations can understand not only how work is structured, but how roles and requirements are evolving.
Complementing our existing global occupation and skills taxonomies, our new tasks taxonomy contains over 37,000 tasks, each of which has been analyzed to estimate how much time people typically spend on it. We have also assessed each task using a framework that considers its variability, complexity, interpersonal requirements, physical demands, and error tolerance, to assess the potential for AI to perform all or part of it.
This helps businesses identify where AI could change how work gets done, and what their people might achieve with the capacity released.
Imagine your business has 100 Software Engineers. Instead of wondering how many of them AI could replace, you can now begin asking a very different set of questions. Which tasks are consuming the most time? Where could AI make the biggest difference? And if we could free up even a small proportion of our engineers' working week, what could we achieve with that additional capacity? Suddenly, AI becomes an opportunity to rethink what your workforce can achieve, rather than simply how much it costs.
Through our task-level analysis of the Software Engineer role, we have identified:
18 core activities associated with the occupation
Time typically spent on each of these activities
Potential for AI automation of each activity
The chart below shows this, firstly looking at general AI potential, and then, when you click on each category, the details of each of the tasks that lie behind these categories. According to our analysis, Software Engineers’ working time is split:
Around 70% on tasks with high AI potential (AI could save more than 50% of time)
26.5% on tasks with low-mid potential (AI can save less than 50% of time)
Under 3% on tasks that are uniquely human (AI cannot save any time)
What does this mean? What it doesn’t mean is that you should replace 70% of your Software Engineers, or even that 70% of your Software Engineers’ working time can be automated, leading to a huge productivity increase.
What it does mean is your company can now have a far more reliable guide to the kinds of tasks that could potentially be automated. So rather than asking how many Software Engineers could be eliminated, you can now understand where to prioritize investments in AI, and how to direct your Software Engineers toward performing other, higher-value activities.
The true value of task-level intelligence is not to enable businesses to eliminate people, but to rethink what their people can achieve. Yet it is only when we add the skills lens that we can fully realize that potential.
Understanding Skills to Unlock Your Workforce’s Potential
We said before that a job is not a bundle of skills, but behind every task there are indeed skills. This means that as AI changes the tasks people perform, the skills required to do these jobs will evolve too. How can we see these changes taking place?
This is where skill-level insights are critical. By monitoring and analyzing skills across the global marketplace, we can see which skills are growing, which are emerging, and which are less prominent.
Looking once again at Software Engineers, our data indicates that the average number of skills requested in job postings for this job increased by 9% between 2022 and 2025. Furthermore, the skills composition of the role is changing. As the chart below, comparing skill demand in 2022 and 2026, shows:
Emerging skills include Artificial Intelligence and Workflow Management
Rising skills include Python, Debugging, APIs, Amazon Web Services, CI/CD, and Kubernetes
Falling skills include Agile Methodology, Java, JavaScript, SQL, Angular, and C#
Stable skills include Software Engineering, Computer Science, Software Development, Git, and C++
The fact that the number of skills being requested by employers hiring Software Engineers is increasing is both fascinating and somewhat counterintuitive. Haven't we just been talking about how AI can automate or augment certain tasks? In which case, why are employers asking for more skills, rather than fewer?
The answer is that work is changing, and so jobs are changing too. But this isn't simply a question of needing more or fewer people to do a particular job. It's also about how the requirements of that job are evolving. Consider this: according to our data, job postings for Software Engineers fell by 46% between 2022 and 2025. Yet over the same period, the average number of skills requested in job postings increased by 9%.
In other words, while employers are hiring fewer Software Engineers, they're asking for a broader range of skills from the people they want to hire.
Bringing Tasks and Skills Together to Drive Growth
Task-level analysis tells us where AI could change the way work gets done. Market-led skills insights tell us how the market is already evolving. But what happens when we put these two concepts together?
Imagine three sports coaches, all of whom have crucial, must-win matches coming up. In the run-up to their games, the first coach pores over matches involving their own team. The second pores over matches involving the opposition. But the third studies both. Which coach has the best insights for the game to come?
The third, of course, because unlike the others, he gets to know the strengths and weaknesses of both his team and the opposition, helping him understand which tactical adjustments to make for the big game.
Joining task-level analysis and market-led skills insights together is analogous to the third coach. A task-level analysis enables you to look inward at your own team’s strengths and weaknesses. Market-led skills insights allow you to look outward at what your competitors are doing. Together, they allow you to see which adjustments you need to make to tasks and skills in your organization, to get the most out of your people and AI. And this is exactly where Lightcast's Work Intelligence comes in, connecting these perspectives to help you make better workforce decisions.
For example, say you wanted to accelerate product development. Task-level analysis can help you identify where AI could free up engineering capacity. Market-led skills insights can help you understand which capabilities are becoming more important and whether your workforce has them. Together, they can help you decide where to invest in AI, where to invest in employee development, and how to redesign work to deliver your business objectives.
In Summary
Returning to the questions we posed at the beginning:
A) How can your organization understand the changes AI is bringing to work?
B) How can you prioritize your response?
A task-level analysis will help you understand which tasks within your workforce AI could automate or augment, and where human input is essential. A market-led skills analysis will enable you to understand how skills are changing across the market, so that you can compare your workforce's capabilities with those your competitors are seeking.
Bring them together, and you have a powerful basis for deciding where to invest in AI, which capabilities to develop in your people, and how to redesign responsibilities around the outcomes your business wants to achieve. Lightcast is uniquely positioned to help you do just that.
Recent Posts

Carpenters’ Salaries in the US: Pay and Purchasing Power by State

Plumbers’ Salaries in the US: Pay and Purchasing Power by State
