For most organisations, the conversation around artificial intelligence still begins in the same place. Which platform should we invest in? How will it integrate with our existing systems? What are the governance implications? How do we manage data security? What productivity gains can we expect?
These are all important questions, but they overlook an important challenge that determines whether AI transformation programmes deliver value or expose an organisation to risk. The greatest barrier to successful AI adoption is rarely the technology itself; it is human psychology.
While boards and leadership teams often frame AI as a technical deployment exercise, with tech and procurement departments tasked with compiling a business case centred around what the tech can do and how much it will cost, employees experience it very differently. Across an organisation, some will see it as a threat, others as an opportunity, experiencing AI as a change to how they work, how they create value, how they are assessed, and in some cases, how they understand their own professional identity.
As AI becomes embedded into everyday workflows, organisations are discovering that technical implementation is often the easy part; human adoption is considerably more complex.
Beyond the technology
Historically, when organisations have introduced new software or systems, the primary concerns were technical. The questions raised have been around correct integration, data migration, and ensuring users know how to use the system.
All of those questions are still valid as part of AI integration programmes, but AI introduces an entirely new layer of considerations, which can determine whether the investment delivers its intended outcomes or creates unintended challenges.
While some employees will be excited by the potential of AI and be curious about how it works, others are much more cautious. My research indicates that people may be concerned that the technology is watching them, or even that, like turkeys voting for Christmas, they may be training AI as their own replacements. This is different and distinct from conventional technostress – the feeling of being overloaded or strained by the pace of technological change. Technostress still exists, but AI is adding a newer layer – an algorithmic anxiety that centres on concerns about whether we are being evaluated by opaque systems, whether our skills remain valuable, and how far data-driven decisions will shape our future.
As humans we are all unique and we want others to see our uniqueness. When AI integration is approached from a purely tech perspective, the algorithmic anxiety of wondering whether decisions are being made about us based on data alone can prevent people from embracing it. In this scenario, not only does the technology fail to deliver its full potential, but the team disengages too.
Unlike previous workplace technologies, AI arrives with a considerable amount of cultural baggage. Long before employees encounter AI in the workplace, they have already formed opinions about it through media coverage, social commentary, popular culture and conversations with peers. Those preconceptions are often positive, with many people approaching AI with excitement or curiosity. But others view it with apprehension or mistrust very few feel neutral.
What ought to be clear is that, regardless of what an AI system can do from a technical standpoint, business leaders should not assume it will deliver those capabilities in their organisation. A sophisticated AI platform that nobody trusts, understands or feels comfortable using will deliver very little benefit. When we talk about human psychology and psychological safety, therefore, it is not simply a wellbeing issue or an issue of engagement and staff retention. It is a fundamental factor in whether AI initiatives achieve the outcomes organisations expect.
Understanding algorithmic anxiety
Algorithmic anxiety is not a single emotion; it is a collection of interconnected concerns that relate specifically to job insecurity and professional identity threat. It often blends classic job insecurity with a more diffuse sense of professional identity threat, as people question how their judgement and experience fit alongside AI systems.
When algorithmic anxiety is in play, team members may avoid using AI tools altogether, or use them only superficially. They may become over-reliant on AI outputs without questioning them and perhaps even feel unable to raise concerns when something appears wrong.
Importantly, these reactions are not necessarily irrational. Often they are a symptom of what happens when an organisation asks its employees to integrate AI into their work without a clear understanding of how decisions are being made, how outputs are generated, or what role humans will continue to play in the process.
Ambiguity and the unknown create incubator conditions for anxiety, so it is no wonder that people presented with something as transformational as AI feel unsure about how much to embrace it, rely on it or reject it. It is the role of leadership to understand that these concerns are not simply resistance to change, but clear signals that people are attempting to make sense of a rapidly evolving environment that has significant implications for their future.
Tackling Algorithmic Anxiety and Leveraging the Early Adopters
A common mistake many organisations make is treating the workforce as a single group moving along a single adoption curve. But that is unlikely to happen, even in tech-led businesses.
People approach AI in very different ways and it is important to avoid assumptions because research, including my own, indicates that common stereotypes do not hold true when it comes to AI adoption. In fact, the commonly held theory that young digital natives are the early adopters has been found to be a red herring in this context, with many young people resistant to AI while their older colleagues are enthusiastic about what it can do.
Rather than relying on a gut feeling about who will embrace and who will reject AI in an organisation, it is important to surface the AI adoption archetypes within the team, and these have been defined as ‘Zoomers’,‘Bloomers’, ‘Gloomers’ and ‘Doomers’. So what are they?
Zoomers are enthusiastic early adopters. They embrace experimentation and tend to focus on opportunities rather than risks. Their challenge is often overconfidence or excessive trust in outputs.
Bloomers are cautiously optimistic. They can become powerful advocates for AI when given appropriate support and guidance.
Gloomers tend to be anxious but persuadable. They want reassurance, transparency and evidence before committing fully.
Doomers are highly sceptical and often resistant. Their concerns may stem from previous experiences of organisational change, distrust of leadership, or broader concerns about technology’s societal impact.
These groups do not map neatly to age groups, seniority levels or technical competence so organisations need to invest in longitudinal research that tracks how people actually feel and develop training that helps them adapt, rather than focusing on technical competency alone.
This is where the involvement of learning and development teams is absolutely vital in changing attitudes and habits, not simply upskilling people to use AI tools. Many organisations respond to AI adoption challenges by investing in training, but low levels of AI adoption are seldom about technical competence. In order to develop the right training and development programmes, L&D teams need to understand concerns about trust, accountability and professional value and surface what support is needed where so that they can address the behavioural challenge.
The hidden cost of AI shaming
One of the risks of failing to understand AI adoption patterns within an organisation is AI shaming, a response that can take multiple forms and leave an organisation exposed to risk.
AI shaming occurs when employees feel embarrassed about using AI or even feel pressure to conceal how much they rely on it due to the attitudes of others.
In practice this could be pointed comments in the workplace, such as ‘Looks like AI wrote this’ or ‘Did AI do that for you?’, with the implicit suggestion that using AI somehow diminishes professional expertise, creativityor effort.
The impact of AI shaming can be surprisingly significant because so many people have built their personal identity and sense of value around their professional knowledge, judgement and experience. If using AI becomes associated with laziness, incompetence or cutting corners, employees may become reluctant to acknowledge how they are using these tools, resulting in a culture of secrecy rather than learning.
It also presents an additional risk because, instead of enabling people to openly discuss AI and how they are using it, people begin experimenting privately. We call this ‘shadow AI and it often emerges when organisations optimise for technical capabilities and control, without investing in understanding who is using AI, how they use it, and what they feel able to disclose.
Shadow AI refers to the unsanctioned use of AI tools outside approved organisational systems. Employees may turn to personal accounts, external applications or specialist tools because approved platforms feel restrictive or poorly suited to their needs. Or they may surreptitiously use personal AI accounts to avoid AI shaming from colleagues.
This creates obvious governance concerns around intellectual property, confidentiality and data security. It also distorts organisational understanding of adoption levels and patterns, suggesting limited use of AI systems on leadership dashboards while employees are actively using the technology elsewhere. When this happens, usage data becomes unreliable and learning opportunities are missed, making risk difficult to spot and even harder to address.
An accurate picture of AI adoption is not only essential for understanding how effectively technology is being used; it is also vital for identifying where systems are adding value, where gaps need to be plugged, who is embracing AI, and how those individuals can be leveraged to support, inspire and develop others. But that accurate picture can only be mapped if leaders create environments where people feel safe to discuss their use of AI rather than hiding it.
From AI deployment to human-AI teaming
So, we understand the human factors that put AI transformation programmes at risk, and we have discussed the AI archetypes that need to be mapped and tracked so that an organisation can understand where targeted behavioural interventions, training and mentoring can be put in place. But how can organisations achieve the cultural shift that converts what they understand about AI’s potential and the obstacles to achieving it into meaningful outcomes?
When organisations talk about AI deployment, the rhetoric suggests a technology project. If we change the language and refer to Human-AI teaming, we change the conversation to one of organisational capability.
This distinction matters because successful adoption requires organisations to think carefully about how humans and AI work together. They can define what remains the responsibility of the human and what should be delegated to the system. They can scrutinise where oversight should sit and how concerns should be escalated.
AI transformation is not a task for the CTO, IT director or procurement team in isolation; it requires collaboration across operations, learning and development, workforce planning, governance, leadership and HR.
A focus on transparency and psychological safety throughout is critical because ambiguity creates anxiety, whereas clarity builds confidence. The organisations that gain the greatest value from AI may not be those with the most advanced technology; they are more likely to be those that create environments where people feel safe to learn. That means encouraging employees to share both successes and failures, making it acceptable to ask questions, and safe to challenge outputs.
Let us not forget that raising concerns about AI is not always resistance. Sometimes sceptics identify genuine risks that enthusiastic adopters overlook. It is not realistic to expect an organisation’s entire team to be enthusiastic adopters. In many cases, the most valuable insights may come from people who are resistant to change and most willing to challenge assumptions or highlight flaws.
Adoption is dynamic and it will be organisations that invest in understanding adoption levels and how they influence behaviours that are most able to leverage the full spectrum of perspectives available to them, enabling their organisation to adapt culturally as well as technically.
AI should never be a one-off implementation programme. By monitoring trust, sentiment, confidence and usage over time, organisations can move beyond measuring whether people are using AI and understand how they are experiencing it, so that they can adapt interventions accordingly. To truly succeed in the age of AI, it is not enough to understand the technology; understanding people is equally critical.
