That’s the question every senior executive should be asking before AI scales across the enterprise.
In December 1984, a Union Carbide plant in Bhopal, India, accidentally released a toxic methyl isocyanate gas, injuring over 500,000 people. Later litigation showed a lack of proper oversight and stewardship by the facility’s management and ownership. History is replete with enterprise examples of the impact of inadequate Stewardship of assets, processes, products, or costs. The Chernobyl disaster in 1986 and the Northeast blackout of 2003 are other examples of inadequate Stewardship.
Not to be overly dramatic, but limited and/or widespread AI-related exposures may be developing within your enterprise. The hidden, underlying cause of this exposure would be a failure to provide adequate Stewardship of AI at the individual, functional, and enterprise levels. The insatiable demand for AI is undermining or delaying AI Stewardship efforts. Some dismiss the need for robust governance as “we have AI governance, controls, or guidelines”, but those are only three elements of commensurate AI Stewardship.
The Hidden Exposure
We all desire the objective of AI to further human flourishing. Every indication is that AI deployment is rapidly cascading into the enterprise at an accelerating pace. Almost all the literature, blogs, news stories, and pundits talk about the now-obvious possibilities and exposures for humanity related to AI. Enterprise senior management must recognize that a lack of enterprise AI Stewardship may create exposures that go well beyond the control of most AI Governance.
Gartner forecasts that the average enterprise will be operating over 150,000 agents by 2030. According to BCG, scale brings its own risks. The more agents a company puts into production, the greater the danger of what BCG calls agent sprawl, with duplicate agents, conflicting permissions, and unclear ownership among other issues. And because agents can combine data, tools, memory, and other agents across workflows, the report warns, risks can “compound at machine speed.”
BCG found that 42% of companies expect AI agents to have real decision-making authority by 2030, but only 5% have enterprise-wide controls to do that safely today. Most companies aren’t ready. BCG just refers to controls. Furthermore, there is the risk of outside, rogue or malicious intent incidents.
AI Stewardship is Broader than Governance
It includes the human capability, management practices, controls, monitoring, resources, accountability, learning, and adaptation needed to put AI to work responsibly and productively. The Hidden Exposure is a lack of comprehensive, formal, disciplined AI Stewardship efforts.
Stewardship involves a variety of players with conflicting, competing, and overlapping roles that must all accommodate a set of shared interests.
We, at Prometheus Endeavor, (PE) believe a more comprehensive and commensurate approach of Balanced Stewardship must be developed over time at the individual and enterprise levels. Otherwise, Stewardship Debt accumulates and senior executives and individuals may very well be responsible at the least, for confusion, out of control costs, operational impacts and possibly a Bhopal or Northeast Blackout level mishap.
Almost every enterprise we examine has a widening gap between the AI being deployed and the Stewardship of the rapidly evolving and changing AI.
Finding a Way Forward
At the very least, Start with a Baseline of Stewardship. Pick a real AI endeavor or set of AI endeavors. Determine the commensurate stewardship required for both the enterprise and individual efforts as they proceed. Then determine what is actually working today. The difference is your stewardship gap.
It would be prudent for every organization to take at least a day and do a commensurate baseline comparing their current needed Stewardship elements, what are in place and how effectively they are actually working. Over time, the fit for your purpose elements should be implemented to match the Stewardship needs.
As AI is rapidly gushing into the enterprise, the AI and how it is being deployed and used is constantly evolving and changing. Thus, an AI Stewardship Baseline should not be a one and done effort. It must be done iteratively and often. The required level of stewardship of course will differ by organization, use case and endeavor so it is critical that the gap assessment is done in light of a specific initiative.
Our Prometheus Endeavor (PE) colleague, John Daniels has created a Guide to AI Stewardship on our PE website. John also provides a more detailed AI Stewardship Assessment Handbook. We hope you find these resources valuable. All we ask is that you honor and acknowledge the attribution to John Daniels and Prometheus Endeavor.
Many enterprises are already behind the curve in AI Stewardship. A process must start with awareness of the need for formal AI Stewardship. Down the road, questions will be asked if an AI effort or series of efforts run off the rails:
- Who will be held accountable when something goes wrong?
- What Stewardship and how well was it implemented did management have in place?
Are you taking the steps to ensure that adequate, commensurate Stewardship is in place or are you leaving yourself and the enterprise exposed?
About Prometheus Endeavor
We are Prometheans — highly opinionated experts, authors, advisors, and educators. We don’t always agree, and that’s the point: conversation sparks insight.
For more than 30 years, we’ve helped organizations and institutions make sense of IT — leading some of the world’s most successful digital transformations. We challenge conventional wisdom, research deeply, ask hard questions, and publish what we learn. Using proven frameworks and decades of management experience, we identify opportunities, avoid pitfalls, and drive meaningful progress.
Prometheus Endeavor is not a consulting firm or attempting to attract revenue. Our Prometheus Endeavor colleague, John Daniels, has developed a publicly available, robust framework for AI Stewardship. His framework has categories for the Institutional domain or level and categories for the individual domain or level. He breaks these categories down into detailed elements.