SHRM26 Debrief · AI, Data & Tech · #2347

Return on Intelligence: Move AI ROI Beyond Cost to Competitive Advantage

with Paul Carney

▶ Watch on the SHRM portal ~87 min, distilled

AI ROIAIQ / AI fluency frameworkchange management

"Trying to use this typical ROI measurements against AI is like trying to determine the value of your cell phone by its weight."

What it was about

Traditional ROI math fails for AI because AI is not a depreciating asset like a building or software license: it learns, compounds, and requires an onboarding investment similar to a new employee. Organizations should replace "return on investment" thinking with "return on intelligence," tracking value across ten dimensions (time, cost, decision quality, speed to insight, capability building, discovery, defensibility, risk management, organizational learning, and confidence) instead of chasing a simple cost-savings number.

By the numbers

only 5% of companies are getting any type of return on investment on it

a report from last August cited on overall corporate AI ROI

10 hours a week down to 90 minutes (saved 8.5 hours/week)

HR leader's manual task time reduction using AI

$300,000 a year down to about $100,000 (saved $200,000)

healthcare company's spend on external recruiters before/after using AI

Key notes

Build AI fluency in strict order: comfortable, then confident, then competent. Skipping the first two steps (as most organizations do by buying a tool and giving one training) is why adoption stalls.
Treat AI like the smartest seven-year-old. It lacks worldly context, is extremely literal about word choice, and will confidently state things that aren't true, so manage it accordingly rather than trusting it like a senior expert.
When AI saves you time or money, don't let the savings quietly disappear back into the budget. Reinvest it visibly (e.g., into a stalled project or new training) and tell that reinvestment story to prove value.

The contrarian takeStandalone AI use policies are counterproductive: because they read as lists of prohibitions, they scare employees away from using AI at all. AI guidance should instead be folded into the existing IT acceptable-use policy. Separately, Paul argues against creating a dedicated "director/head of AI transformation" role, comparing it to the DEI-hire pattern that got cut a few years later, and says AI accountability should sit with the whole executive team rather than one appointed AI leader.

Take this back Monday

Do this for your team

Pick one AI-saved time block this month and publicly reinvest it (a stalled project, extra training) instead of letting it vanish into budget.

Say this in your next leadership meeting

AI ROI isn't a depreciating-asset calculation, it's more like onboarding a new hire, so we're tracking value across ten dimensions, not just cost savings.

Watch out for

Jumping straight to the "competence" stage (buying a tool, giving one training session) without first building comfort and confidence with AI.
Measuring AI adoption with simple compliance metrics (e.g., logins, mandatory training completion): employees will game the metric (KPMG example: logging in and out just to register a session) while gaining no real value.
Treating AI ROI like ROI on a building or software license, a fixed, amortizable asset, when AI is dynamic, compounding, and closer to onboarding a new employee than installing a tool.

Fun fact · Paul Carney

The former Chief HR Officer turned AI educator and his wife have earned a Level 2 Award in Wine from the Wine & Spirit Education Trust.

If this landed, see these

↳ Go deeperStrategic AI Adoption: Designing Smarter WorkflowsTurns the comfortable-confident-competent fluency ladder into an actual process-by-process rollout plan.⇄ The counterpointHR Executive Insights: A CHRO's View on Owning the Digital Workforce, Operationalizing Generative and Agentic AI, and the Future of WorkBets on mandated play and micro pain-point fixes over a ten-dimension measurement framework.✦ The unexpected oneChange-Ready or Change-Resistant? How to Assess and Elevate the Mindsets That Drive AgilityThe comfort-before-competence ladder for AI maps onto this session's vertical-development model for change readiness.