“Did we just light $1.2M on fire?” Some version of this question is going on right now in most business leaders’ heads as they grapple with low returns on AI investments.
These leaders are walking an almost paralyzing tightrope right now. C-suites and boards have sky-high expectations about investing in AI to radically improve productivity, efficiency, and performance. Yet almost every project is failing to deliver the expected results, as shown in a recent MIT study that showed a 95% failure rate on AI implementations.
So what are leaders supposed to do?
This is the prime moment for HR and L&D leaders to step in. Not as AI technologists (although bonus points if you have AI expertise!), but as the architects of human systems that determine whether AI adoption in the workplace succeeds or stalls.
The MIT study further showed that the top barriers to AI were largely human, with only one of the top challenges being about the AI model itself.
It’s also worth noting that the issue isn’t an overall resistance to AI. In fact, the study found that many employees were using “shadow AI,” meaning that they were using AI tools of their choice even if it wasn’t the tools the company was telling them to use.
So the issue isn’t just the technology. Rather, it’s how leaders are handling the change.
In the rush to get AI solutions off the ground, companies are creating an emotional overload for people. People who feel overloaded shift into threat response, narrow their focus, and protect what they know. Adoption, experimentation, and learning all crater.
If you want AI to work at scale, you have to expand emotional capacity before you increase technical complexity. Because even a simple solution that is adopted and integrated into the work of people will deliver far more returns than an advanced technical solution that isn’t applied.
That’s the job of emotionally intelligent leadership, especially frontline managers.
The biggest danger right now is to look at the 95% failures as proof that it’s not going to work. Instead, we should be looking at the 5% that are doing it right, because they are going to quickly outpace their competitors.
The outliers that move from pilot to P&L don’t necessarily have fancier models of AI. They lead differently in three practical ways:
You don’t need an AI lab to fix this. Let’s look at a few practical strategies you can use to help your managers implement emotional readiness for AI.
Treat BEATS as leading indicators that predict your lagging KPIs (time‑to‑productivity, cycle time, CSAT, cost‑to‑serve).
Before go‑live, require: BEATS at/above target for the impacted groups; managers able to name top fears and how they’ll be addressed; talking points for “what this means for my role.” If the gate isn’t passed, remediate, then proceed. Slower for a week, faster for the quarter.
Sequence capacity before complexity. Coach managers to: label emotions in the moment; reframe demands; restore a sense of control (small choices, quick wins); and model curiosity over certainty. Dealing with uncertainty is now THE skill to master.
Don’t dump an internal marketing campaign and pray. Pair rollouts with facilitated sessions that connect skills to this quarter’s priorities and today’s use‑cases. Practice the hard conversations (“Will this replace parts of my job?”) in the room.
Stand up lightweight listening posts with employees, customers, and risk/compliance. Validate concerns publicly, change designs visibly, and close the loop. That’s how you earn the trust to scale.
Days 0–30: Baseline & listen. Add BEATS to weekly team check‑ins; stand up manager office hours in the most‑affected workflows.
Days 31–60: Build capacity. Run a Resilience‑Builder Manager sprint (two 2‑hour sessions) with coaching in the flow of work; equip managers with conversation guides to build resilience in others and detect dips in BEATS metrics.
Days 61–90: Pilot with guardrails. Launch a scoped AI pilot behind an Emotional Readiness Gate, pair it with manager support tools, and scale only if BEATS rise and KPIs move.
Most AI programs aren’t failing because the tech is weak. They’re failing because people are exhausted and managers aren’t equipped to lead through it.
HR and learning teams stand at a crucial moment to integrate emotional intelligence their operating system—measure BEATS, train manager readiness, and build real feedback loops. If they do, AI adoption in the workplace moves out of pilot purgatory and into business results.