The Human Capability Preservation Problem: What Skills Must We Protect in the Age of AI?

16 Sep, 2026

Enterprise AI Perspectives, September 2026

As AI becomes increasingly embedded in workplaces, education, and decision-making, a new question is emerging:

What human capabilities are we at risk of losing, and which ones must we actively preserve? That question was at the centre of the inaugural Enterprise AI Perspectives event hosted by the UQ Centre for Enterprise AI last week: The Human Capability Preservation Problem.

Moderated by Professor Tim Miller, the discussion featured UQ researchers Dr Sarah Bentley, Associate Professor Ida Asadi Someh, and Professor Deb Brown, who each brought distinct perspectives from social psychology, business information systems, and philosophy and critical thinking.

To frame the discussion, each panellist selected a video that explored a different dimension of the human capability preservation problem. Together, they promoted discussion on agency, expertise, critical thinking, trust, and organisational capability in an increasingly AI-enabled world.

Why This Matters for Enterprise AI

For organisations pursuing enterprise AI adoption, success will not be determined solely by the technology they deploy. It will also depend on the capabilities of the people that use these tools. However, research suggests that, if not designed and used properly, AI tools have the ability to erose these very capabilities.

As AI becomes embedded into workflows, organisations face a dual challenge: capturing the benefits of automation while ensuring employees continue to develop judgement, expertise, creativity, and critical thinking. Enterprise AI therefore presents not only a technology transformation challenge, but a human capability transformation challenge.

Beyond Efficiency: The Importance of Human Agency

Drawing on her selected videos, Dr Sarah Bentley explored the importance of maintaining human agency when working with intelligent systems. While AI can reduce routine work and support productivity, she cautioned against the hyperbole surrounding these tools and the risk of learned helplessness, where people gradually cede cognitive power, in this case, to technology. A key message was that people should work from a position of power and agency, with AI supporting rather than directing their thinking.

Another key point was that organisations are only as capable as their people, and the knowledge, expertise and judgement held by individuals and teams remain critical organisational assets. The challenge is not simply adopting new AI tools, but ensuring people continue to develop and exercise the capabilities that create value.

The discussion also challenged the assumption that removing the drudgery of routine work automatically leads to better outcomes. While AI can increase efficiency, some forms of expertise, judgement, and creativity are developed through immersion in processes or work that organisations may be tempted to automate. The question is not simply what tasks can be delegated to AI, but which experiences contribute to the development of human capability.

The Risk of Skill Erosion

Associate Professor Ida Someh’s video choice prompted discussion about the impact of AI on organisational expertise and decision-making. As a case study researcher, Ida often examines extreme cases to reveal broader patterns and emerging risks. One such risk is what happens when organisations hand too much decision-making authority to AI systems.

While automated decision support can improve efficient and scale, there will inevitably be situations where a person needs to intervene, question or override and AI-generated recommendation. If humans are removed from the decision-making process for too long, they may no longer possess the judgement and expertise developed through practice to do so effectively.

Rather than focusing solely on AI adoption metrics, Ida challenges organisations to ask more strategic questions around organisational capability including: What skills do we want to preserve for the future? How do we measure whether those capabilities are strengthening or eroding over time? And how do we connect capability preservation to organisational value and return on investment? As AI becomes increasingly embedded in workflows, maintaining critical knowledge and expertise is just as important as achieving productivity gains.

These questions are especially important for enterprise leaders seeking to realise long-term value from AI investments. Productivity gains may be realised quickly, but capability erosion often emerges slowly and can remain hidden until organisations face novel situations, unexpected disruptions, or decisions that require human expertise. Understanding this balance represents an important frontier for enterprise AI research and practice.

Critical Thinking in the Age of AI

Professor Deb Brown’s selected video focused attention on critical thinking as one of the most important and transferable human capabilities in the AI era. Drawing on work from UQ’s Critical Thinking Project (UQCTP), she highlighted that critical thinking remains one of the most in-demand yet most deficient skills in the workforce. The proliferation of large language models has only heightened this. Developing both the cognitive skills and metacognitive awareness needed to evaluate information, question assumptions, and reflect on reasoning is therefore becoming increasingly important.

Deb distinguished between cognitive offloading, where technology supports thinking and decision-making, and cognitive surrender, where people become reliant on AI for tasks, they have never learned to do themselves. The challenge is understanding when offloading is helpful and when it risks diminishing capability. She argued that people need to think critically to engage critically with AI, and that this foundation must be developed before individuals can effectively evaluate, challenge, and interpret AI-generated outputs.

The discussion extended beyond the workplace to education. If people are to engage critically with AI, they must first have opportunities to develop the cognitive and metacognitive capabilities that underpin good reasoning, judgement, and decision-making.

For organisations, critical thinking is a strategic capability. As AI-generated content, recommendations, and decisions are becoming more commonplace, the ability of employees to evaluate evidence, challenge assumptions, and exercise sound judgement may increasingly differentiate organisations that thrive from those that become overly dependent on technology.

Trust, Truth, and Accountability

A recurring theme throughout the discussion was the relationship between AI, trust, and truth. While AI systems can provide powerful support for analysis and decision-making, they do not inherently distinguish between what is true and what is merely plausible.

The panel emphasised that critical engagement remains essential. Questioning outputs, challenging assumptions, understanding reasoning processes, and maintaining accountability for decisions will continue to be fundamentally human responsibilities. As AI capabilities advance, preserving these behaviours is just as important as preserving technical expertise.

Connect with Us

The UQ Centre for Enterprise AI offers a unique opportunity for organisations to collaborate with world-class researchers and thought leaders to shape the future of artificial intelligence in enterprise settings. Our partnerships are designed to deliver mutual value—advancing your strategic goals while contributing to responsible, impactful AI innovation.

Professor Tim Miller timothy.miller@uq.edu.au

Professor Shazia Sadiq s.sadiq@uq.edu.au