20 Examples Of Non-Inclusive Behaviours In An AI-Enabled Workplace

What happens when old workplace biases meet new AI technologies? Here are 20 examples of non-inclusive behaviours.

AI is revolutionising our workplaces at a pace we are struggling to keep up with. But coming at the same time and speed are new ways of excluding people, which means we have to pay attention to manage those behaviours effectively. If we don’t, we are turning the clocks back on the steps we made towards inclusive workplaces before the days when we relied on technology-assisted processes and systems.

AI is changing how organisations recruit, manage performance, allocate work, monitor employees and make decisions. But as we are finding out, technology does not automatically make workplaces more objective or inclusive. AI can reproduce existing bias, introduce new forms of exclusion and make discriminatory decisions harder to see.

Non-inclusive behaviour in an AI-enabled workplace therefore isn’t limited to what people say or do. It can also be embedded in the processes and technologies organisations choose to use and how they go about implementation. .

20 non-inclusive behaviours in an AI enabled workplace

1. Dismissing concerns

Labelling employees as “resistant to change” when they may have legitimate concerns about privacy, job security, accuracy or bias. Viewing them as “luddites” and responding with the “system is neutral” line rather than listening and investigating their worries, especially if a person works in a sector or function which is high risk for layoffs.

2. Inadequate procurement protocols

Poor procurement analysis of AI tools can harm employees by introducing biased software, violating privacy, creating unfair workloads, and excluding workers with disabilities. It can also mean that, if believing as the customer, you are outsourcing the risk, you may not be, and without in-depth analysis may leave the organisation open to legal penalties.

3. No policy or governance systems in place

When companies deploy AI without strict policies, they essentially outsource management decisions to unvetted mathematical models, leaving employees exposed and vulnerable to bias.

4. Inadequate human oversight of AI systems

When AI systems operate without meaningful human oversight, a problem often called “automation bias” occurs when organisations treat algorithmic outputs as absolute truth. This transforms AI from a helpful tool into an unchallengeable manager, stripping employees of agency and nuance.  If you top it off by blaming the algorithm and absolving yourself of accountability, then that is a leadership cop-out for sure.

5. Unequal access to AI training

Giving some employees extensive AI upskilling while others are left behind creates a new workplace skills divide. This is especially true for those directly impacted by AI implementation initiatives. Excluding employees from decisions about AI that affect their work and introducing AI systems that change jobs, monitoring, or performance expectations without meaningful employee consultation also creates a skill and knowledge divide.

6. Failure to audit for bias

Assuming an AI-generated recommendation is objective simply because it is produced by technology. AI is riddled with bias and needs to be audited regularly and benchmarked against existing or target data. This can be in talent management processes, recruitment, performance reviews, and promotion recommendations. In Europe under the EU AI Act, the penalties for user organisations for not doing this are potentially severe.  The process has been outsourced, but the risk hasn’t. In the US multiple class actions have been filed against vendors too.

Must Read: DE&I 2.0: Why DEI Must Evolve with AI

7. Using AI-generated recruitment criteria without validation

Adopting AI recommendations about “ideal candidates” without checking whether they reproduce historical patterns of exclusion. AI results should be benchmarked on a skill-based assessment required for the role, not on historical real-life bias.

8. Penalising employees who do not use AI

Making assumptions that employees who are reluctant to use AI reflect poor attitude or low skill levels penalises individuals who may lack access, have different working styles, or have accessibility needs. It may not indicate poor performance. Ultimately, these non-inclusive behaviours and assumptions end up damaging trust and embedding any concerns even deeper.

9. Using AI surveillance disproportionately

Applying productivity monitoring or behavioural analytics in ways that disproportionately affect particular groups or create a climate of fear. This leads to fear-based anticipatory trauma as employees develop anxiety about being laid off or put on a performance improvement plan.

10. Using AI to analyse employee sentiment without transparency

Collecting or analysing employee communications or behavioural data without clearly explaining what is being monitored and why. This can be emails, Slack messages, or any other internal communication noting language that might indicate positive or negative sentiment, which damages trust, interferes with honest feedback and communication. It may even break GDPR if not done openly with very explicit information about what the data will be used for.

11. Accepting AI-generated performance feedback without checks

Delivering generic or inaccurate feedback because it was produced by an AI tool rather than based on meaningful human observation. There are a number of well-documented hazards around:

• AI hallucinations when tools frequently invent false details, misattribute projects, or misremember specific metrics when synthesising large data pools.
• AI slop: Automated text relies on predictable patterns, producing vague platitudes like “Displays strong leadership skills” without citing actual instances of impact.
• The data blindspot: AI only knows what is logged in system software and misses vital in-person and offline contributions such as mentoring colleagues, de-escalating team conflicts, or steadying morale during a crisis.
• The bias multiplier: Large language models can mirror historical workplace biases. They routinely use softer, passive language for female employees while using assertive, execution-oriented terms for male employees.
• The erosion of trust: Employees instantly recognise robotic, generic text. Receiving a machine-written review signals to a worker that their manager does not value them enough to write a personalized assessment.

12.  Allowing AI to influence promotion oppportunities

Using AU to influence promotion decisions without human oversight can be excluding.  AI models learn user preferences of the company leaders and exhibit proxy bias of those who use the tools and might end up favour some groups over others based on historic pattetns or previous promotions related  gender, race, or age. Recommendations can also be based on old or inaccurate data and fail to provide accurate reasons forfor being promoted or otherwise.

14. Accepting  gender stereotyped recommendations

AI has been notorious for gender stereotype assumptions in certain areas, especially those that have been traditionally considered male and female areas of expertise.  Leadership, technical or financial ability might be associated with men. Women may be assigned carer and support roles, portrayed as mothers, etc. Research shows that if women use AI for salary negotiation advice, then they are advised to settle for a lower salary than if the prompt indicated they were male.

15. Ignoring accessibility when introducing AI tools

Introducing systems that are difficult or impossible for employees with different accessibility needs to use effectively. It’s important to factor in accommodations when assessing implementation programmes to avoid discrimination. Examples of inaccessible and disability-excluding AI-based tools are unfortunately not uncommon: for example, a speech recognition system not able to understand commands made by a person with Down syndrome or a speech impediment. In automated interviews, candidates with visual impairment who are not able to sustain eye contact with the camera may be penalised.

Important resource:  Trauma Informed Interviewing Techniques – 3Plus International

16. Using AI to generate workplace communications without human review

Sending insensitive, culturally inappropriate or exclusionary messages because “AI wrote it” is unacceptable adn potentially damaging. Just because the process is automated, it doesn’t mean that there should be no human oversight.

17. Using AI-generated images that reinforce stereotypes

Allowing employer branding, recruitment campaigns or internal communications to repeatedly depict particular jobs, roles or leadership characteristics through gendered, ableist or racial stereotypes.

18. Failing to challenge AI-generated misinformation about an employee

Treating an AI-generated summary, profile or allegation as fact without checking its accuracy. AI even tells you to check with a waiver at the bottom of any content. That is one things you should trust about it.

19. Using AI-assisted complaints without safeguarding the complainant

Treating an AI-written grievance as less credible, less authentic, or less worthy of investigation. Any complaint, whether written by AI or an individual, should be given full respect and complete analysis. AI-assisted complaints are likely to be more sophisticated, which means that HR and any investigators need to dive deeper to establish the facts without retraumatising the complainant.

20. Using AI to intensify workplace harassment

Technology Facilitated Gender Based Violence is on the rise and entering our workplaces. This includes creating or circulating deepfakes, manipulated images, synthetic sexual content, or targeted harassment involving colleagues. HR and DEI professionals plus line managers need to be brought up to speed on these developments. Abuse no longer walks through a door but appears on a screen.

When we try to make AI-enabled workplaces as inclusive as possible being intentional from the outset with an inclusive mindset will be vital to prevent regression. It won’t be enough to carry on regardless and try and fix things later.

Is your workplace ready for the inclusion challenges AI creates?

AI may be changing how we work, but responsibility for fairness, inclusion and accountability still sits with people. Organisations need to ensure that AI tools, policies and workplace practices don’t reinforce existing bias or create new forms of exclusion.

3Plus International helps organisations build inclusive, AI-ready workplaces through practical training, workshops and consultancy.

Talk to us about creating a more inclusive workplace in the age of AI.