Needed: An Executive ‘Off Switch’ For AI When Things Go Wrong

Is there an ‘off switch’ for AI when we need it? It all boils down to governance and accountability. Unfortunately, there’s just too little of it, as AI continues to inundate workplaces and lives.

Consider this: only 13% of organizations have the governance foundations they need to move into AI in a safe and meaningful way, a recent survey of 2,100 executives out of SmarterX shows. The percentage with an actual roadmap is only slightly better – 29%. A third of respondents, 32%, report their organization have no defining frameworks, such as gen AI policies, ethics policies, or an oversight body.

“AI doesn’t run itself," states the report, authored by Tayloe Radey, director of research for SmarterX. "Someone has to lead and decide which tools the organization uses, how they’re deployed, who gets trained, and what guardrails are in place. Without that ownership, even the best intentions around AI tend to stall or go sideways.”

About 59% leave AI decisions to their executive leadership, and 31% assign it to their technology leaders. Sixteen percent report no clear owner or are not sure who is making their AI decisions.

The ability to assign accountability is, unfortunately, a critical weakness seen in AI-saturated organizations these days, agreed Maribel Lopez of Lopez Research, who has been studying these developments and talking to business leaders about these issues.

Business leaders – whether or not they are directly immersed in AI in their companies – need, at a minimum, veto power over machine decisions, she explained in a recent webcast hosted by Jo Peterson, CIO of Clarify360.

AI governance practices are too far and between, and where they do exist, they are a hit-or-miss proposition scattered across departments such as legal and HR. Lopez found in her research. Add to the challenge a lack of skills to understand and keep AI on the right track.