Carolyn Geason-Beissel/MIT SMR
Not a day goes by without another article being published about how AI could disrupt yet another aspect of our business or personal lives. In recent years, AI adoption has indeed taken off. However, if you pay close attention, you’ll notice a dichotomy.
Many examples of successful early adoption of artificial intelligence tend to come from a small cluster of industries that are heavily digitized or are pro-technology. The usual suspects include banking, financial services, e-commerce retailers, and the like. However, some other industrial sectors, many of which are big contributors to our economy, don’t show the same level of progress or enthusiasm when it comes to AI adoption.
Take the example of specialty and essential services industries such as construction, mining, or waste management. Some of these companies are part of a robust economy but largely powered by legacy software from decades ago, with some processes still happening through pen and paper. While AI has made nascent inroads here, the levels of adoption leave much room for growth.
Leaders in these industries often assume that they have stable processes that have served them well for decades. Yes, things might break once in a while, leading to customer service disruption, rework for the team, and internal process disruption. But then, they have always recovered. People in these industries may view AI as gimmicky, too much work, and/or not trustworthy.
Having spent more than 15 years helping dozens of industries embrace AI, I’ve been curious to study what distinguishes the two sets of leaders and the quite different levels of AI adoption they achieve. And, importantly, I’ve spent years in the trenches experimenting with techniques that help address adoption challenges.
Here, I’ll share what’s at the root of the leadership challenge and how leaders in industries that have been conservative about AI can orchestrate meaningful change. Let’s examine some grounded examples and no-nonsense tips for AI adoption.
Why AI Adoption Lags in Some Industries
My experience in the field points to three prevalent factors holding back some industries from moving forward with artificial intelligence.
1. AI feels inaccessible and scary.
When you can’t comprehend something, you start developing a fear of it. When everyone around you seems to talk about it and you feel left behind, the fear only grows. When the technology feels intrusive and uncomfortable, you draw back into your shell.
This is exactly what’s happening with AI when it comes to a majority of late adopters in both private and public sectors. The hype around AI and the seemingly irrational excitement of tech pundits only alienates people in cautious companies. To make matters worse, anytime there’s news about an uninformed AI investment backfiring or machine learning algorithms going rogue, it solidifies the narrative that AI is inaccessible and not ready for the masses yet.
Driver-facing AI-enabled cameras in freight vehicles are a case in point. For truck drivers, a camera inside the cab feels intrusive and disciplinary long before it’s perceived as a safety or performance-aiding tool. A report by the American Transportation Research Institute shows that truck drivers’ approval of driver-facing cameras tends to be low: just 2.24, on average, on a 0-to-10 scale among 650 current users from across the industry.
2. AI looks like a lot of avoidable work.
AI is often touted as a savior that automates drudgery. But people on the ground who are tasked with making the AI tools work and integrating them into workflows may perceive AI as creating extra work, not relieving them of it.
With front-line teams in labor-intensive industries often feeling overstretched and under-supported, the need for more training or changes to existing workflows just adds friction before adding any value. In many late-adopting industries, AI is immediately associated with capital-heavy hardware and forced operational change.
It doesn’t help that organizational memories are often clouded by many failed or painfully stretched technology rollouts — think enterprise resource planning systems, safety tools, telematics systems, and so on. People wonder whether this AI-tools wave is another fad that’s worth waiting out. When you take a deeper look, you realize that change fatigue, not an aversion to technology, is the real blocker.
3. AI benefits don’t really seem worth the pain.
Most technology evangelists and leaders commit the blunder of communicating AI value in the wrong currency. Improved accuracy or productivity boosts mean little to front-line operators, who care more about customer escalations, rework, or operating costs.
In a 2025 executive survey by Deloitte, although 65% of leaders said that AI is part of their corporate strategy, many also acknowledged that the ROI is neither immediate nor purely financial. From a front-line worker perspective, the cost of learning and adopting an intimidating technology like AI feels personal, but the benefits feel abstract and impersonal.
When it’s difficult to articulate tangible business outcomes from AI for the next quarter, such initiatives struggle to secure or sustain sponsorship and are easily deprioritized. Every time AI implementations fail to deliver on vague goals, which is quite often, the trust deficit only grows.
Three Pillars for Successful AI Adoption
How can you, as a leader, address those challenges and set your organization up for success? Consider these three essential strategies.
1. Use everyday analogies to make AI less threatening.
Education is a prerequisite for meaningful AI adoption. When your end users don’t understand why they should use or trust AI, the initiative is dead on arrival. How can you make AI accessible to an audience that’s not digital-native?
We are no longer in a period where there are few notable uses of AI. Some people don’t realize that they already use AI dozens of times every single day. Don’t we unlock phones with facial recognition? Aren’t even unbranded smartwatches good at detecting workout activities or flagging an irregular heart rhythm? Don’t some people delight at discovering long-lost school buddies through Facebook or Instagram friend recommendations?
Each of these examples is an instance of AI at work. In conversations with leaders, when I share these as examples of sophisticated AI use by the general public, it surprises them every single time. Once the technology is reframed this way, conversations can begin to shift from fear of AI to a curiosity around where else it might be at play. You make real progress when you demystify AI through familiar experiences rather than technical lectures.
This framing also enables a more honest discussion about the potential of AI and the threat to jobs. In many professions, people then begin to appreciate that they are more likely to lose opportunities not to the AI itself but to other humans who know how to use AI better. This strengthens AI’s positioning as assistive and AI tool use as another skill to acquire.
Take the case of AI platform Hey Bubba, designed for trucking owner-operators and small trucking companies. Instead of using dashboards or complex workflows, the system operates entirely through voice. Drivers can search and book freight, negotiate with brokers, find parking, and book hotels through natural conversations, with the help of AI. This service works because it builds on familiar uses of AI assistants, such as Siri and Alexa, and thus feels natural.
2. Integrate AI into systems people already use.
Is it easier to renovate a house or ask people to move into a brand-new one with unfamiliar rooms, rules, and routines? With AI adoption, you want to take the renovation approach. It’s a blunder to try a big-bang approach to roll AI into an organization.
Always start with incremental changes to existing workflows and software. Remember that your teams already use dozens of software tools. These are the best starting points where leaders can inject AI and gently nudge user adoption.
For example, most front-line teams already live inside software, such as billing systems, customer relationship management systems, dispatch tools, maintenance software, or safety logs. Some of these systems may be clunky, but they are heavily used and largely unavoidable. The pain points within these systems could act as perfect entry points to introducing AI — places where users could see the value and welcome the initiative with open arms. When AI meets people where they already work, curiosity replaces resistance.
Take the case of fleet maintenance. Most technicians and supervisors already spend their days inside a computerized maintenance management system. Work orders are logged there. Inspections are recorded there. Breakdowns are investigated there.
An effective approach to introducing AI that can predict vehicle failures, for example, is to embed AI directly into the maintenance systems users already trust. AI can flag recurring fault codes, highlight assets with rising failure risk, or suggest prioritizing certain work orders before a breakdown occurs.
3. Quantify AI’s impact using metrics people already track.
Once you make AI accessible and identify familiar avenues to inject it, the quickest way to earn buy-in is to lead with the business result it unlocks.
Start by anchoring AI value to outcomes that stakeholders really care about and are judged on. Usually, there are two perspectives: creating upside (growth or throughput) or preventing downside (lost revenue or risk reduction). Examples of upside metrics are win rates, or asset utilization, while downside metrics include cost leakage or service disruptions. Remember: New KPIs always trigger debate and delay action, whereas familiar metrics accelerate alignment.
Next, pick a combination of short-term impact and long-horizon projections. Sticking just to lag metrics could disillusion stakeholders, who need to see quicker momentum to retain confidence and excitement for AI. For example, reduction in customer complaints is an example of a lead metric to validate short-term progress, while incremental revenue from repeat customers is a lag metric that might need a few quarters to start materializing.
Consider the example of an industrial materials distributor focused on accelerating growth. The company struggled to systematically identify and act on new business opportunities. Field sellers relied on manual, time-intensive methods, such as driving through cities to visually spot new construction projects. The process was inconsistent, slow, and difficult to scale.
The company built an AI engine that combined internal sales data with external signals to score and prioritize potential opportunities and recommend relevant products. Generative AI was then applied to extract insights from unstructured public data, such as construction permits, to identify upcoming capital projects.
These insights were embedded into existing sales workflows to personalize outreach at scale. The approach unlocked new opportunities in the first year, significantly expanding the sales pipeline and improving success rates for email outreach — both of which were existing sales metrics that stakeholders already cared about.
Where AI Adoption Is Really Won or Lost
In late-adopting industries, AI doesn’t fail because the technology falls short. AI often fails because leaders underestimate the human and operational context in which AI tools are introduced. We must remember that front-line skepticism is not resistance to progress — it’s just a rational human response that can be influenced when tackled strategically.
The organizations that move fastest follow a clear progression. They demystify AI by promoting understanding among people; embed AI into existing workflows before forcing new ones; and prove AI’s value using metrics that are already being used to reward or penalize people. When these conditions are met, adoption becomes a pull factor as opposed to a hard push.
The way forward for late-adopter industries is not to imitate tech-first sectors but to adopt AI on their own terms. Successful leaders treat AI as a capability to be woven incrementally into daily work rather than a system to be rolled out abruptly. In these environments, user comfort and trust, not algorithms, ultimately determine whether AI delivers on its promise.