The AI ROI trap: Measuring the wrong things is masking AI's real costs

AI advisor and UN speaker Edosa Odaro founded The AI Values Institute to break out of the dominant narrative, which assumes “more capable AI automatically creates more value.”

The reality is that “AI does not automatically create value,” Odaro tells UNLEASH.

This explains why enterprises still struggle to reap ROI from AI; MIT data concluded that 95% of organizations reap zero returns from their AI investments, while Gartner found only one in five organizations has seen significant or transformative value from AI.

The issue, according to Odaro, is many companies have fallen into the trap of prioritizing  short-term financial returns while neglecting larger, longer-term impacts on society.

The AI Values Institute’s mission is to flip the decision-making script, by proving that enduring financial value from AI is only possible if AI progress includes and benefits everyone, not just the few.

To achieve this shift, Odaro tells UNLEASH organizational decision-making on AI needs to shift from “How can quickly can we adopt AI?” to “How do we create AI value that remains measurable, trusted and sustainable over time?”

As The AI Values Institute launches its inaugural whitepaper, UNLEASH spoke with Odaro, and two other founding members of the Institute: Professor Amy Shi-Nash from the AI Institute of Monash University and Professor Markus Krebsz of the UN Economic Commission for Europe - Risk Management and Market Surveillance.

They share the mistakes companies are making around AI value, what leading companies are doing differently, and where HR fits into the equation.

Allie Nawrat: Why is focusing purely on short-term financial ROI from AI an expensive mistake for employers to make? What are the risks to their business, employees and wider society?

Edosa Odaro: Organizations that optimize only short-term financial outcomes often discover they've unintentionally weakened the very conditions that allow value to scale.

One of the biggest misconceptions is that financial value and human value compete — in reality, employees determine adoption, customers determine trust, regulators determine legitimacy, and society determines an organization's licence to innovate.

Markus Krebsz: An organization treating AI as an efficiency lever while ignoring its human dimension is accumulating hidden liabilities.

The ethical ROI (namely reputational capital, regulatory goodwill, and employee confidence) is invisible in a traditional business case but the cost of losing it arrives fast.

Labor displacement is a board-level strategic governance question, not a transformation program issue to be handled below the line.

Organizations that automate faster than their workforce can retrain face crystallizing regulatory enforcement and evidenced litigation risk.

A social license to operate, as well as trust, is extraordinarily expensive to rebuild once lost.

Amy Shi-Nash: Companies fail because they lack clear intentionality — they rush into technology testing without defining what they actually want to achieve, and end up with narrow, ineffective outcomes.

Only prioritizing financial gains leads to cultural degradation, lost talent, and public backlash, especially as societal pushback against AI-driven job displacement grows.

AN: If they’re not focusing on financial metrics like productivity and efficiency, how should organizations measure AI success?

MK: Productivity and efficiency metrics tell you whether the AI system is doing what it was asked to do; they say nothing about whether that is safe, fair, or sustainable. Genuine AI governance needs a richer measurement set: Key X Indicators covering performance, risk, and control together.

Specifically, this includes bias differential across demographic groups, hallucination rates, human override frequency, and foundation model version tracking.

AS-N: Long term success of AI can’t solely be predicated on efficiency and productivity.

It must include a view of how this powerful technology drives creativity, new market opportunities, and employee empowerment.

EO: AI success should be measured across multiple dimensions, including business performance, decision quality, workforce adoption, customer trust, governance maturity and societal outcomes.

Sustainable AI leadership require measuring whether AI is strengthening the organisation as a whole, rather than improving isolated metrics.

AN: What are companies ahead of the curve on AI value doing differently?

EO: The organizations pulling ahead treat AI as an enterprise-wide transformation, not a collection of technology projects — competitive advantage increasingly comes not from deploying the most AI, but from creating AI value that's sustained, evidenced and trusted.

MK: The pattern among organizations genuinely ahead of the curve is behavioural rather than sectoral: they treat AI governance as a board-level strategic capability, not a compliance cost.

The GDPR precedent is instructive here – organizations that invested in genuine data governance before May 2018 consistently outperformed those that scrambled to the minimum, in both compliance costs and customer trust metrics.

What they do differently is specific: named individuals accountable for every deployed system, pre-deployment safety cases completed before go-live rather than filed retrospectively, and continuous monitoring throughout the lifecycle rather than monthly or quarterly narrative reports to committees.

AN: As organizations reframe AI value, where must HR sit in the conversation? What are their roles and responsibilities?

MK: HR sits at the intersection of two of the most consequential AI governance dimensions: workforce capability and conduct risk. 

AI literacy is not a one-off compliance exercise; it decays without sustained investment.

HR is best positioned to design, maintain, and measure it across the organization: from board-level strategic literacy to first-line awareness of what should and should not be inputted into an AI tool.

EO: HR has become central to AI value creation because many AI failures are not technology failures; they are alignment failures. AI changes how people work, how decisions are made and how organisations evolve.

HR therefore plays a critical role in building capability, trust, adoption and organisational resilience alongside AI and technology implementations.