AI’s ROI Paradox: Beyond Technology

Most businesses struggle to achieve AI ROI despite high adoption, hindered by insufficient governance, skill deficits, and a human bias towards less reliable, human-like AI. True potential lies in addressing these human-centric challenges beyond technology.

The grand promise of artificial intelligence, a technological revolution poised to reshape industries and boost productivity, is increasingly being met with a rather stark reality: for many businesses, the much-hyped ROI remains elusive.

A recent MIT study painted a sobering picture, suggesting that a staggering 95% of enterprise AI implementations have been, by and large, fruitless.

This isn’t merely a minor speed bump on the road to innovation; it’s a fundamental disconnect between aspiration and outcome, a chasm that a new study suggests is rooted in something far more intrinsic and, ironically, more human than we might expect.

One might assume that a technology failing to deliver would see a dip in adoption.

Yet, the findings from a global survey by data analytics company SAS and the International Data Corporation (IDC) reveal a curious paradox.

Despite the widespread struggle to monetize AI, businesses are embracing it at an astonishing pace.

Well over half of the respondents (65%) are already using AI in some capacity, with an additional 32% planning to do so within the next year.

This aggressive push into AI, however, is taking place against a backdrop of profound, yet often unacknowledged, distrust in the very tools being deployed.

The study points to a critical misalignment: a significant majority of IT professionals and business leaders (78%) claim to have “complete trust in AI,” yet a paltry 40% have actually bothered to implement the fundamental governance and explainability guardrails necessary to ensure their AI systems are, in fact, trustworthy.

This isn’t just an oversight; it’s a strategic blind spot.

How can one claim complete trust in a system without establishing the mechanisms to verify its reliability, fairness, and transparency?

This gap between declared confidence and demonstrable control is, according to IDC’s Chris Marshall, precisely why AI’s potential remains largely untapped and ROI suffers.

It’s a classic case of hoping for the best while failing to prepare for the worst, or even the merely mediocre.

The impediments to achieving genuine trust and, consequently, tangible ROI, are distilled into three primary roadblocks.

First, weak cloud infrastructure often hobbles AI capabilities from the outset, a technical hurdle that can sometimes be outsourced or upgraded.

Second, insufficient governance stands as a significant barrier, a systemic failing to establish clear rules, ethical guidelines, and oversight for AI deployment.

This isn’t just about compliance; it’s about building a foundation of accountability.

Third, and perhaps most critically, is a pervasive lack of AI-specific skills among the existing workforce.

While the first two can often be addressed with technology or policy, the skills gap speaks to a deeper human investment problem.

Thankfully, the data suggests that business leaders are prioritizing training initiatives over knee-jerk layoffs, recognizing that the human element remains irreplaceable, even as roles evolve.

Adding just one AI-related skill, it seems, can significantly boost an employee’s market value – a silver lining in a rapidly changing landscape.

But the SAS-IDC study unearths an even more intriguing, and perhaps unsettling, psychological phenomenon at play: our inherent bias towards AI that feels human.

The survey revealed that respondents tended to place significantly more trust in generative AI systems – like ChatGPT, Gemini, or Claude – than in older, more transparent machine learning models.

This is counter-intuitive.

Traditional models are built with fewer parameters, making their decision-making processes relatively easier to understand.

Generative AI, by contrast, is a black box, notoriously opaque and prone to “hallucinations” or fabricating information.

Why this preference for the less reliable?

The study authors suggest it’s a deeply ingrained human quirk: we reflexively trust systems that mimic human interaction.

Generative AI excels at producing human-like language, creating an illusion of understanding, empathy, or even consciousness.

This illusion, however benign it might seem, can have profound implications.

It can lead users to form emotional attachments to chatbots or “AI companions,” imbuing these algorithms with an undeserved aura of authority and intelligence.

“The more ‘human’ an AI feels, the more we trust it, regardless of its actual reliability,” the authors conclude.

This observation cuts to the heart of the matter: our emotional and psychological responses to technology are shaping its adoption and perceived value, often overriding logical assessments of its actual trustworthiness and utility.

Ultimately, the struggle for AI ROI isn’t solely a technical problem; it’s a complex interplay of inadequate infrastructure, governance failures, skill deficits, and a powerful, often subconscious, human bias.

To truly unlock AI’s potential, businesses must look beyond the shiny new tools and confront these deeper, more human-centric challenges.

It requires not just investing in smarter machines, but in smarter, more discerning, and better-prepared people, all operating within a framework of robust, transparent governance.

Only then can the promise of AI truly begin to deliver on its extraordinary potential, moving beyond the realm of costly experimentation to become a genuine driver of progress.