In The Week’s “Deskilling: A Dangerous Side Effect of AI Use,” the first line beneath the image reads:
“Atrophied skills have been observed across a wide array of fields, including medical and mental health.”
As an organizational learning and development strategist for a large behavioral health care provider, I have had this on my radar for quite some time, but I believe it should be on yours as well.
Even more, as a longtime L&D professional and computer information systems educator with professional and academic backgrounds spanning adult education, higher education, information technology and organizational learning, I find this outcome unsurprising.
Of course, technology has always reshaped work. What is different today is the speed with which large language models have been introduced at scale, often without sufficient preparation, governance or thoughtful integration. Hype cycles have a way of running ahead of practice, and eventually the dust settles.
The Week’s article also provided examples from each of the three primary domains in which I work:
- In mental health, therapists are “allowing themselves to become passive in the act of therapy,” essentially becoming a “supervisor over the AI use for therapy” while limiting their “reflexive diagnostic thinking,” according to Forbes.
- In technology, computer coding is increasingly being delegated to AI, shifting human developers toward integration, monitoring and higher-level analysis, according to the American Enterprise Institute.
- In education, many students are using AI to write essays or conduct research. Yet, as The Atlantic notes, “The term paper, for all its tedium, teaches a discipline that’s hard to reproduce in conversation: building an argument step by step, weighing evidence, organizing material, honing a voice.”
Over the past three years, I have intentionally observed this unfold. I’ve followed analyses from fellow L&D professionals and from professionals across industries. Some of those conversations have lacked nuance, particularly in sectors like health care, where the context differs significantly from many commercial environments. Too often in these conversations, AI has been presented as inherently good and urgently in need of implementation without sufficient clarity about the problems being solved, the value being created or how people will adopt, integrate and sustain the change over time.
For nearly my entire career, I have worked to understand both sides: the human and the technological. I have never chosen a lane. But I have always believed there is an ethical order of prioritization.
One of The Week article’s most important observations is that deskilling is not inherently negative. It has occurred throughout history and has contributed to many of the advances we benefit from today. The more important question is not whether deskilling occurs, but, rather, how much is appropriate, in what contexts and with what safeguards. Those questions require systematic and systemic thinking.
The article also highlights another concern I have also been raising: What happens when the technology is unavailable?
When I recently discussed on LinkedIn the importance of surgeons maintaining the ability to perform procedures without robotic assistance should a system fail mid-operation, another LinkedIn user made a similar point. The same principle explains why pilots continue to practice manual landings even though commercial aviation relies heavily on automation. Automation can extend human capability, but resilience depends on preserving essential human expertise.
Then there is the question of equity. At times, conversations on LinkedIn proceed as though everyone, everywhere, has equal access to the same technologies and will experience AI in the same way. They will not.
Not every health care system can afford multimillion-dollar surgical robotics platforms. When I underwent surgery in 2013, the robotic system being used at Johns Hopkins Medicine was, at the time, unique within my region. Access to advanced technology varies considerably across health care organizations, communities and countries.
We also tend to overlook the ongoing costs of these systems. Sometimes discussions treat AI adoption as though it is inevitable and self-sustaining. Yet these technologies depend upon continual investment in fresh knowledge and research, much of which comes from institutions that some now argue AI is replacing, including higher education and journalism. They also depend upon significant computing infrastructure, energy, water and the people who build, maintain, govern and improve these systems.
Because I completed my doctoral work within a school of education and urban studies grounded in a social justice mission, equity will always be one of the lenses I bring to both research and practice.
That perspective also shaped my own doctoral research. I examined how systems usability varied across regions, including rural and urban communities where differences in digital infrastructure significantly affected access and user experience. Those same infrastructure differences continue to matter in health care today.
I also tried to ensure my professional experiences reflected those realities. In 2012, I accepted a significant pay cut to work in a rural community while completing my doctorate. Later, I returned to a larger suburban region outside Washington, D.C., continuing my work across different organizational settings. Experiencing multiple contexts has been important to developing a broader perspective rather than assuming one environment represents everyone.
I mention this because it matters. When people make decisions that affect others, they must understand the people those decisions affect and make reasonable efforts to understand their experiences. Not everyone will have opportunities to work across as many settings as I have, but I have found one principle to be consistently true: When we neglect those whose needs are less visible, the consequences eventually extend far beyond those communities.
This is why I continue to advocate for partnership, unhurried practice and thoughtful exploration before deciding what should and should not be delegated to AI.
Those decisions should be considered through multiple lenses. Philosophically, we should ask questions about ethics, equity, justice, human agency, professional responsibility and what kinds of expertise we believe society ought to preserve. Practically, we should examine patient safety, workforce capability, organizational resilience, regulatory requirements, cost, infrastructure, accessibility, continuity during system failures and long-term learning.
Technology has always changed what people do. The more enduring question is whether we are being equally intentional about preserving what people still need to know.
