AI in the workplace is inevitable – but are we using it responsibly?

Whether we welcome it or not, the use of artificial intelligence (AI) in the workplace is reshaping how organisations hire, operate and evolve. A report by Stanford University has highlighted just how uneven that transformation may be, stating AI’s ‘significant and disproportionate impact’ on entry-level roles in the US. Despite overall employment remaining stable, early career opportunities in AI-exposed fields are declining.

Similarly, in the UK, graduates are facing the toughest job market since 2018 due to employers pausing hiring and leveraging AI to reduce costs. Couple the declining market with the fact that graduates from low socio-economic backgrounds are 32% less likely to be offered entry-level professional roles than more advantaged applicants, and the gaps of inequity begin to widen and deepen.

Organisational leaders are responsible for determining the strategy and action plan relating to the deployment of AI. It is thus interesting that AI is often referred to in a way that suggests it is a sentient being, with ‘free will’. The accelerating pace of AI is directly linked to both the breadth of AI access – for example, its embedding in search engines – its deployment within organisations, and the extent to which employees utilise AI to augment and then transform how the business of work is executed. History has shown the importance of IT in enabling organisations to maintain competitive advantage, and for the public and third sectors to continually seek insights that enable them to deliver services that improve the quality of life for the people and communities they serve.

While AI continues to accelerate workplace transformation, it is not acting on its own. The technology – and how it is utilised – reflects the decisions, priorities and existing biases of the people and organisations that design and adopt it. The fundamental question for employers, therefore, is not whether AI should be used, but whether it is being used responsibly.

AI does not make the decisions – humans do

AI is a tool with extraordinary transformational aspects, but we should consciously question, is it really AI that is transforming the workplace and impacting the workforce, or human/leadership decisions that are shaping the change and the outcomes being sought? As human beings we all have unconscious bias, which is an innate element in us all. We produce the systemic conditions, products designs, codes written and data sources corralled. It is therefore critical that we consciously consider the fairness and accuracy of the information that is delivered to the end user, intervening where necessary to minimise unconscious bias.

AI does not make isolated decisions; it identifies patterns in the data it has been trained on and applies those patterns consistently. While we know that AI is constantly evolving, we should question to what extent it is aiding or hardwiring previous decisions and assumptions. When used in recruitment, for example, if a candidate’s profile is considered a poor fit, the system is likely to treat similar profiles in the same way.

As a result, we start seeing historical patterns – such as who gets hired and who gets promoted – encoded into systems that perpetuate those biases at scale, and constantly so. People may apply to an organisation for several roles, especially if it is an organisation they admire. Have we stopped to consider that once a profile is overlooked or rejected, it may continue to be overlooked or rejected, thus inadvertently reducing the talent pool that we are accessing?

This is where the impact of unconscious bias shifts from being individual to systemic. What might once have been a subjective human judgement becomes an automated pattern applied across thousands of candidates, and because the assumption is that systems are objective, people are less likely to question their outputs.

Yet, AI cannot assess what often matters most in human potential. As Donna Herdsman, Freelance Management Consultant and DEI Advisor, formerly of PwC, IBM, Hewlett Packard Enterprise and Korn Ferry, explains: “The devil is in the detail. Only the hiring manager knows exactly what it is they're looking for. It’s going to be a mixture of the candidate’s experience, attitude and their potential to thrive in an organisation, which is something AI cannot truly or accurately measure.”

This is why organisations must remain consciously inclusive when designing and adopting AI tools. Without considered human oversight, the utilisation of AI risks becoming a gatekeeper and reinforcing the experience of exclusion faced by minoritised groups, becoming a bias perpetuator, rather than an enabler for broadening accessibility.

Automation, augmentation, and applying critical thinking

Almost half of UK workers are worried that AI will take their jobs, according to a GMB Union report. A large part of this anxiety comes from the idea of AI automating – and therefore replacing – jobs. Organisations often use AI automation to reduce costs, increase efficiency and scale. Examples include AI chatbots replacing customer service agents, or AI systems reviewing CVs to screen and include or rule out candidates. But what happens when automation is deployed without meaningful human oversight?

Automation has been around for a long time, but what's more recent is the growing tendency to trust AI without the application of critical thinking – potentially treating AI as the ‘voice of ultimate truth’. We must consider the risk of assuming that something must be true because AI says it is. Curiosity, therefore, becomes a key skill requiring consistency of application if equity and fairness are to be achieved.

Human judgement remains a priority, which aligns with the concept of augmentation. Augmentation, in this instance, refers to AI technology enhancing human work, assisting employees to perform more efficiently, supporting decision-making and boosting productivity. For example, instead of automatically screening candidates, augmented AI might summarise a candidate’s job profile for the person who is leading the recruitment to review.

In fact, China is accelerating workplace AI adoption, supported by national AI policies, while “stepping up investment in human capital” and aiming to “build a workforce better equipped” for rapid technological change, suggesting a shift towards AI augmentation rather than pure automation. The key takeaway is that AI should enhance human capability, not replace it. Organisations that invest in upskilling employees to work alongside AI are more likely to equip their workforce for the future.

Examples of AI automation versus AI

Explore five examples of AI automation versus augmentation in the workplace.

Using AI while maintaining the ethics of diversity, equity and inclusion

AI systems are only as objective as the data they learn from – and that data reflects the world not only as it exists today, but also the commentaries and views expressed over time, depending on the knowledge cut-off and whether real-time data is accessed. For example, an analysis of commercial face-analysis software found that darker-skinned faces were "misclassified" at a much higher rate than any other group, due to biased training data, insufficient representation, and technical limitations in how facial features were detected and interpreted. The consequence of this is exclusion, not just inaccuracy.

Being aware of bias is not enough; organisations need to become consciously aware and proactively manage the risks arising by actively investing in systemic inclusion redesign. Such redesign could include diversifying datasets when training AI systems, establishing frameworks and guardrails within AI systems to mitigate bias, or designating a diverse panel of employees to review and challenge AI outputs.

Democratise AI or risk creating a two-tier workforce

Accessibility is one of the biggest inclusion risks surrounding AI. As the technology becomes a part of everyday work, access to AI is quickly becoming a workplace equity issue. Employees who regularly use AI are more likely to develop new skills, work more efficiently and build confidence with emerging technologies. Those without access or official training may unintentionally be left behind, as organisations struggle with proving the ROI of AI adoption, addressing privacy concerns, or establishing clear guardrails.

These challenges have led 71% of UK employees to turn to unapproved personal AI tools (Microsoft data), widening the gap between those who are building AI skills and those who are not. Furthermore, 60% of professionals say they would be more likely to use AI at work if proper training were available, according to a Henley Business School report.

Additionally, the World Economic Forum reports that women have been slower to adopt AI tools, raising concerns that existing inequalities could be reinforced if access, training and support are not provided equitably and tailored to the needs of the individual. Organisations need to give access to everyone and help them get comfortable utilising AI, otherwise they risk creating a two-tier employee experience – not because of capability, but because opportunities to develop and apply AI literacy can influence visibility, performance, progression and promotion of talent within an organisation.

5 ways to adopt AI responsibly through conscious inclusion

Organisations that fail to embrace AI risk undermining productivity and losing their competitive edge. The challenge is ensuring AI promotes inclusion rather than unintentionally scaling exclusion.

AI is neither inherently objective nor inherently biased – it simply reflects the decisions, values and data provided by the people who design and use it. If organisations fail to challenge AI outputs, diversify its datasets, or provide their people with equitable access to AI tools and training, leaders risk perpetuating existing workplace inequalities at an unprecedented scale.

For CEOs, HR and DEI leaders, responsible AI adoption means embedding conscious inclusion throughout. Here are five ways to adopt AI in the workplace responsibly to ensure parity, fairness and inclusive equity:

1. Establish clear AI governance and guardrails. Develop company-wide principles that define where AI can and cannot be used, who remains accountable for decisions, and how AI outputs should be reviewed. Build bias mitigation, privacy, transparency and ethical considerations into these frameworks from the beginning.
2. Keep humans accountable for high-impact decisions. AI should support decision-making, but not replace it. Recruitment, promotion, performance management and disciplinary decisions should always involve meaningful human oversight and application of critical thinking.
3. Provide equitable access to AI tools and training. Democratise AI by ensuring all employees have access to approved tools and learning opportunities. Avoid creating a two-tier workforce where only some employees develop AI literacy and have enhanced career advantages as a result.
4. Audit AI systems regularly for bias. Regularly review AI outputs to identify unintended patterns of discrimination or exclusion. Use diverse reviewers, challenge assumptions and continually refine systems to improve fairness and accuracy.
5. Maintain curiosity and critical thinking. Encourage employees to challenge AI outputs, fact-check information and understand the technology's limitations. Remember that AI can hallucinate (provide content that appears factual but is ungrounded), generate outdated information, and perpetuate historical bias if its outputs are taken as fact and not critically reviewed.

As Donna explains, we’ve faced and embraced more change than we give ourselves credit for. If we stop and think, is AI simply not the next evolutionary wave? The organisations that succeed won't be the ones that fear AI, but the ones that ensure everyone has the opportunity to benefit from it. Ultimately, AI will not determine whether workplaces become more inclusive – organisations will. It's not AI but us humans who decide how to use AI to the benefit, or otherwise, of us all.

For more information, register for our webinar, AI in the Workplace: How to Adopt it Responsibly and Reduce Hidden Bias.

Additionally, explore our conscious inclusion programme, which helps organisations build the confidence and capability to challenge bias, break down invisible barriers and work towards closing the gap of inequity.