Change Lessons From GenAi’s Failure Rate

If you’ve been working in the talent development and talent management space over the last few years, you have not been able to escape the all-encompassing shadow of generative AI. Generative AI has been the talk of the talent development world, whether it’s a report from Deloitte about AI’s impact on headcount or its potential use in HR investigations. You cannot attend a conference or listen to a talk about the field without someone mentioning the impact of AI. Thus, the MIT report on the high failure rate of AI interventions is one that all talent development professionals should learn from.

This report shows 95 percent of AI interventions deliver no discernible impact. One of the key factors highlighted in the report was the desire of some companies to build their own internal AI tool. While we might think that the problem is that the AI tool isn’t strong enough in terms of its data, the reason these researchers have found for the failure is an implementation one. In other words, traditional change management. Rather than technological, what we find is a failure of vision and leadership.

The researchers found that targeted implementation of AI was much more effective than mere adoption. This is similar to the suggestions that my business partner, Dr. Mike Chetta, and I made in our chapter on boutique consulting. We create a framework to help those considering boutique consulting in ATD’s Handbook for Consultants. We suggest that you should be clear about what you have to offer. In the same way, any intervention requires this level of clarity in its implementation.

Let’s reframe these AI interventions as traditional interventions, whether a leadership development program, a new hiring system, or some other talent management exercise. At the most basic level, we’d follow change management best practices. At Talent Metrics Consulting, we use the 5 Es to manage change:

• Equip – Get your organization ready to make the change. Especially from a cultural perspective, the organization must be ready to accept the new change and the new culture that comes along with it.

• Envision – Create the project goals for the change and a plan for the change process involving KPIs and key stakeholders.

• Executive – At this point, you should start your actual change, whatever that change may be.

• Embed – Embed the changes within the processes of the organization through rule changes, process changes, and reward systems.

• Evaluate – Evaluate the effectiveness of the change and implement further changes to achieve the goals set in stage 1.

There are two major lessons from these AI implementations.

1. Shorting their preparation. Many organizations are not spending enough time in the preparation phase and developing a change plan. Rather than seeing how AI fits into the organization, they decide that they must become involved in AI and implement the intervention without a clear vision for how AI will work.

2. Buying rather than building. Many of the interventions that worked for the organizations involved external vendors. This may seem counterintuitive, but these vendors offered very specific products with specific outcomes. Your organization may decide to build an intervention, but it may be easier from an implementation and service standpoint to partner with a vendor that offers a specific solution.

Before jumping feet first into any project or change, make sure to use the five Es to ensure a successful change process. When in doubt, do something targeted with a trusted consulting partner, and you’re more likely to have a successful project.