How businesses can get ahead of potential AI deskilling

What is AI deskilling, and why is it a problem?

AI deskilling is the process by which humans who adopt and depend on AI tools become less capable of thinking critically or solving complex problems.

A primary cause of AI deskilling is the ability of generative and agentic AI tools to make processes feel cognitively easier by automating cognition-intensive tasks -- like writing code or documents -- that traditionally required extensive thought.

As a result, some workers might begin to pull back from performing cognitively intense tasks, such as reasoning through novel challenges, believing that heavy thinking is no longer part of their jobs.

Deskilling vs. job displacement

Don't conflate AI deskilling with AI-driven job displacement. The latter term refers to the wholesale outsourcing of work to AI for roles no longer require human workers. In contrast, deskilling primes workers to stop thinking of cognitively heavy tasks as parts of their job, even when those tasks are necessary to solve challenges that AI can't handle.

Consider an entry-level software development job versus a senior software architect role. The former job has the potential to be displaced by AI tools, which are adept at writing basic code. But a senior architect's work involves making nuanced decisions about aligning software system design with business needs, security and compliance priorities, and performance goals. It can't be fully displaced by AI.

Nonetheless, a senior architect who becomes accustomed to relying on AI to outsource routine tasks, such as generating boilerplate code or mapping architectural patterns within a microservices application, might become so comfortable offloading work to AI that the architect no longer thinks critically about application design. As a result, the architect might unquestioningly accept an AI tool's architectural recommendations without performing the critical thinking necessary to validate AI-generated output.

How AI deskilling affects businesses and roles

Using AI to accelerate tasks that AI tools can handle effectively is beneficial for businesses. Allowing workers' critical-thinking skills to atrophy due to AI deskilling is not.

Consider how AI deskilling can affect the following roles:

• Software developers. When using AI heavily, developers could become less capable of thinking critically and solving problems in novel ways, which could lead to missed opportunities for optimizing code or architecture.
• Marketers. When using AI to help plan campaigns and content, marketers might think less creatively, resulting in marketing initiatives that aren't tightly aligned with target audiences or business messaging goals.
• Sales teams. When using AI extensively, sales associates risk failing to think and operate in the nuanced ways necessary to build trust with high-value clients and navigate complex sales cycles.

Shortcomings in areas such as these can have direct, immediate consequences for businesses. They could reduce the ROI of investments in areas like engineering and marketing. Meanwhile, less effective sales teams could result in decreased revenue across the board.

Furthermore, the deskilling problem isn't just limited to business contexts. Students using AI also correlate with reduced critical thinking skills, according to MIT research. For businesses, the tendency toward AI deskilling among workers might begin before they reach the workplace, making it even more urgent to get ahead of the challenge by developing anti-deskilling strategies.

Long-term AI deskilling impacts: Business risks CIOs can't ignore

The risks of AI deskilling don't end with less effective operations or revenue reduction in the near term. There are also likely to be long-term business consequences:

• Increased operational, compliance and security risks. Workers who unquestioningly trust AI output place the business at risk. AI tool mistakes or inaccuracies could disrupt business processes. They could also trigger compliance and security risks if, for example, an AI tool shares sensitive data with a party that shouldn't have access to it.
• Erosion of a business's innovation capacity. When workers no longer think critically, it becomes much harder to gain a competitive edge by designing and building the best products and services.
• Talent pipeline degradation. Businesses that lose their competitive edge might have a harder time recruiting talented workers over the long term. Employees who are primed to be true innovators won't want to work in environments where AI bots do most of the "thinking."
• Overreliance on AI tools and vendors. When workers become so dependent on AI tools that they can't make decisions without their help, the business ends up beholden to those tools -- and the vendors who sell them -- in ways that could stifle innovation. For example, a company might not be able to migrate to an alternative AI platform if its employees are deeply invested in using the current one, especially if the workforce has lost the innovation capabilities necessary to evaluate and implement alternative technologies.

It remains to be seen whether AI deskilling will result in business challenges like these. To date, researchers have only been able to measure the short-term impacts of AI deskilling, since the phenomenon has been around for just a few years. But it's not hard to imagine the long-term business impacts.

Balancing AI deskilling losses with efficiency gains

For business leaders focused on the bottom line, accepting some level of AI deskilling might seem appropriate if the tradeoff is a massive boost in efficiency. For instance, allowing a software engineering team to become 50% less innovative could seem worth it if the team can bring apps and features to market 10 times faster with AI tools.

Arguably, however, this type of tradeoff is a recipe for long-term business failure. Efficiency and productivity boosts are not a substitute for innovation; at a certain point, it doesn't matter how quickly teams can produce outputs if the outputs are too low in quality or unoriginal to create a competitive edge. Releasing software 10 times faster only benefits the business if the code is innovative enough that people actually want to use it.

Therefore, measuring AI deskilling's effects and its relationship to efficiency gains requires looking beyond basic metrics, such as how quickly workers complete tasks. It also demands a subjective analysis of what innovation looks like for a particular business and which skills workers must retain to drive innovation.

Innovation capabilities are hard to track quantitatively. AI's effect on an organization could appear positive if leaders focus only on productivity metrics. But if businesses don't strategically assess the effects of AI deskilling risk, they might be unaware they're losing their innovation capabilities until it's too late.

How to prevent AI deskilling while scaling AI adoption

The answer for most businesses is not to avoid AI adoption. On the contrary, AI can deliver massive efficiency gains, and it would be unwise to miss out on that opportunity in the interest of avoiding the pitfalls of deskilling.

Business and technology leaders must find ways to encourage workers to use AI where appropriate, while also ensuring they retain the willingness and ability to think critically when it matters. Effective tactics include the following:

1. Define explicit human roles within AI workflows. Designating exactly when and how humans should contribute to AI-driven processes -- such as validating an AI agent's recommendations before they're applied -- helps workers recognize and value their role in AI-centric operations. It also emphasizes the critical thinking skills that only humans can contribute.
2. Require workers to document and explain AI outputs. When workers are asked to validate AI outputs by explaining in depth what the outputs entail and the pros and cons they offer, it prompts critical thinking. It also avoids scenarios where validation turns into pro forma signoffs by workers.

3. Designate AI-free processes. Identifying certain work that employees must perform without the assistance of AI solves two challenges: It ensures that AI doesn't play a role in processes where it's not appropriate, and it explicitly requires human workers to perform cognitively intense tasks as part of their jobs.
4. Quantify innovation abilities. While it can be difficult to quantify innovation in concrete terms, businesses can examine data on how the effectiveness of marketing campaigns or the efficiency of software applications correlates with AI use. When AI yields less effective outcomes, the business can use these insights to show workers how overreliance on AI undermines organizational goals.
5. Support and reward critical thinking. Practices that incentivize employees to think critically -- such as bug bounty programs for programmers who discover bugs in AI-generated code -- demonstrate that the organization values cognitively intensive work. So does compensating workers for completing technical training or certification programs for skills that could be partly offloaded to AI. This practice underscores that the organization wants its workers to be skilled in these areas, even if AI can handle some of them as well.

Chris Tozzi is a freelance writer, research adviser, and professor of IT and society who has previously worked as a journalist and Linux systems administrator.