The AI Inclusion Challenge: Can Technology Make the Workplace More Inclusive?
AI can make workplaces more inclusive, but only when we make deliberate choices about how it is introduced and used. That means giving people access, training and time to learn, recognizing different needs, and ensuring that decisions affecting employees remain fair and accountable. The real measure of success is whether AI helps more people contribute, grow and be recognized for their capabilities.
The honest starting point is that AI does not make a workplace more inclusive on its own. Without deliberate attention, it can widen the very gaps we hope it will close.
The people who pick up a new tool first are usually the ones who already have an advantage – technical confidence, some spare time, and a manager who treats experimenting as real work. Everyone else waits to be told it applies to them. A year later you have two populations inside the same company, and the gap between them is wider than the one you started with. Nobody plans for this. It just happens if you leave it alone.
AI can also remove some barriers that existed long before generative AI arrived. It can help someone communicate more confidently in a language that is not their first, make information easier to nderstand, support different ways of learning, and give an employee outside the technology function access to capabilities that once required specialist help. That is real inclusion. But even here, the tool only creates the possibility. If people are uncomfortable using it, do not have access to it, or fear being judged for needing it, the barrier has simply moved rather than disappeared.
This broadens inclusion beyond functions and technical confidence. The same tool will not work equally well for everyone. An employee with an impairment may need it to integrate with assistive technologies. Someone working in a second language may need multilingual support. Others may need more time and guidance to get comfortable. Asking employees what helps and what gets in their way is part of choosing and introducing the tool, not something to leave until later.
So, the question is not really whether AI can make a workplace more inclusive. It is whether we design for it. And inclusion here is not a programme with a name. It is a series of fairly ordinary decisions about access, time and opportunity, taken deliberately instead of by default.
At Excelsoft, the first of those was access. We made AI tools available across the organisation rather than only to the technology teams, and support functions were part of that from the start.
The second was training across all of those functions, rather than assuming the technical teams would work it out and everyone else would follow. People learn a tool when someone shows them what it does for the work they actually do. A finance executive needs to understand the basics of AI, including its limitations and how to use it safely. But the training becomes meaningful when it connects to the work they recognise – such as a reconciliation they run every month.
The third was to sort out governance and security centrally, before rolling anything out. This matters more than it sounds. In many organisations, AI use is not necessarily restricted; it is simply difficult to access. You need approval, it is not clear who gives it, and the request takes time. The people who chase it are usually the ones who are already confident and well connected. Everyone else gives up quietly. So, this is not an administrative step that comes before inclusion. It decides who gets to use AI at all. The same goes for the things nobody writes down: whether a manager treats time spent experimenting as work, whether support functions are asked before tools are chosen, and whose ideas get taken forward.
We have also made AI adoption part of KRAs and KPIs across roles. This was deliberate. If learning a new way of working is not counted anywhere, it becomes something people are expected to do on top of their real job. But expectations need to fit the role and the support available. The aim should be useful, responsible application, not simply more usage. Knowing when AI is unsuitable is also part of learning to use it well.
Open forums are one way of testing whether any of it is working. We ran an AI ideathon, open to everyone, at any grade, in any function. About seventy-five ideas came in – from market intelligence, presales, sales and business development, not only from the technology teams. Ideas from sales and business development won, and one of them is now going into our hackathon to be prototyped and tested. That taught us something we had not quite expected. Understanding of a customer problem is spread far more widely across an organisation than the org chart suggests.
For a long time, acting on that understanding required someone who could build software, which meant it travelled through the technology team, or it did not travel at all. When the tool stops being specialist, the advantage shifts to whoever understands the problem best. Often that is not the technology team. Which is why AI literacy should not sit only with technology functions: keep it narrow and you keep the benefit narrow, and you lose the ideas of the people closest to the customer.
I should be clear about where we are. This is early. Opening something to everyone is not the same as everyone taking part – the people with more time, more confidence and more encouragement from their manager still tend to move first. Counting ideas is easy. The harder question is who is not participating, and why, and whether it is the same set of people each time. Rather than assuming openness alone resolves the issue, we are reviewing such cases. The next step is to speak to those who are not taking part: do they need more time, a more relevant example, or encouragement from their manager? Their answers should shape the support that follows.
Much of the current discussion treats AI skills as a technical qualification – something you either have or need to acquire. In practice what matters more is adaptability: the willingness to try a tool, work out where it helps and where it misleads, and change how you work. That is not a technical trait, and it is not distributed along the lines people assume.
As AI takes over more routine work, human skills become even more valuable. Judgment about when the output is wrong, communication with a customer who is unhappy, and the ability to ask whether the question itself was the right one all matter more rather than less. These are not lesser skills or a fallback for people who are not technical; they are what the work increasingly depends on.
There is another side to inclusion: how AI is used in decisions about people. When it helps shortlist candidates, assess performance or recommend development opportunities, organisations need to ask whether its recommendations are fair and what they might be missing. Someone must remain accountable for the decision, employees’ information must be protected, and people should have a clear way to question an outcome. Easier decision-making is not automatically fairer decision-making.
For HR, this is a more substantial role than it might first appear. It is not about policing tool usage or writing an AI policy and filing it. It is about making sure access, capability and opportunity are spread evenly rather than settling with the same people they usually settle with – the confident, the technical, the ones who are already visible. That means asking who has the tools, who has had the training, who is being given time to experiment, and whose ideas are actually getting heard. Those are people questions, and they decide whether AI adoption narrows the gaps inside an organisation or widens them.
The principles are straightforward. Putting them into practice takes deliberate choices and sustained attention.
Technology does not make a workplace more inclusive. Decisions do – about who gets access, who gets taught, and who gets a fair chance to show what they can do. Get that right and the technology will surface capability you did not know you had. Get them wrong and it will simply reward the people who were already doing well.
For me, the future of high-stakes assessment is not about replacing paper with screens or adding more AI. It is about creating an examination system that gives candidates a fair opportunity, works reliably across different conditions, protects the integrity of the process, and keeps people accountable for important decisions. As AI becomes more capable, we will also need to rethink what we test-placing greater emphasis on how candidates apply knowledge, reason through problems and demonstrate real capability. In the end, the real test of any assessment system is whether people can trust the result. For further insights into the evolving workplace paradigm, visit
About Author
Shruti Sudhanva, CPO Excelsoft technologies
Shruthi Sudhanva, Whole‑Time Director and Chief People Officer at Excelsoft, is redefining HR as a core driver of business performance. Her priorities workforce planning, productivity, and capability building are powered by AI to align talent strategy with business outcomes. Focused on selective hiring, retention, and continuous learning, she embeds DEI into processes and builds leadership pipelines, especially for women, while driving AI‑led productivity that blends technology with human creativity and disciplined execution.
- The AI Inclusion Challenge: Can Technology Make the Workplace More Inclusive? - September 22, 2026
- How Superworks is Helping CHROs Navigate India’s HR Digitization Shift - September 21, 2026
- Sagar Pandey on Winning the Talent War in BFSI Recruitment, Retention & Employee Experience - September 9, 2026
