On the MIT Study “The GenAI Divide: State of AI in Business 2025”
- Catherine Louropoulou
- 5 days ago
- 3 min read
According to this, 95% of organizations do not reap significant returns on their investments in Generative Artificial Intelligence (GenAI). But what do the data behind the numbers show?
We have seen a great deal of commentary in both the Greek and international press about this study, which so far has been used more as clickbait than as a serious source of facts about the realities and prospects of Artificial Intelligence (AI).
The heightened attention to its findings is also linked to the soaring stock valuations of tech giants. The report presented a notably cautious — even negative — picture of Generative Artificial Intelligence (GenAI). According to its authors, “95% of organizations gain no meaningful return” from related investments, despite the explosive rise in U.S. tech stock capitalization in recent months.
But how accurate is that claim? Let’s take a closer look.
Key Findings
Broad adoption, low returns:
Despite massive investments (estimated at $30–40 billion), 95% of organizations fail to achieve measurable financial gains from GenAI.
The “pilot project” gap:
Only 5% of initiatives manage to move from the “pilot” stage to actual production deployment.
Integration challenges:
The main barriers are organizational rather than technological — for example, how AI is integrated into existing workflows, how staff are trained, how value is measured, and how “one-off experiments” are avoided.
Market contradiction:
While the shares of major tech companies (Microsoft, Nvidia, Google, etc.) have skyrocketed, the real business value for most firms has yet to materialize.
A Closer Look at the Study’s Parameters
Definition of “return” or “value”:
The study defines “return” as a measurable impact on financial results (P&L) within a short timeframe (e.g., six months after a pilot).This definition excludes other forms of value — such as productivity gains, process efficiency, or strategic benefits — which are harder to quantify but can be very significant.
Methodology and sample:
The research is based on over 300 publicly known initiatives, 52 structured interviews, and 153 executive surveys. However, even the report notes that these interviews are “directionally accurate” rather than fully validated corporate data.
Time horizon:
Findings are based on short-term trends — some projects may yield returns over a longer period than the six months examined.
Undervalued indirect benefits:
Critics argue that the report overlooks improvements in quality, speed, decision-making, reputation, and innovation — all valuable but non-financial outcomes.
The risk of generalization / clickbait headlines:
The “95% failure” figure makes for an eye-catching headline but can be misleading if interpreted without context.
Corporate diversity:
Not all organizations are the same — size, industry, infrastructure, AI maturity, and culture significantly influence outcomes.
Our Perspective
In reality, the “95%” figure does not mean that AI is failing — it means that most organizations have yet to see direct financial returns because the technology is still in an early adoption phase.
Moreover, many of AI’s most valuable effects — improved productivity, better customer experience, fewer errors — are not easily captured in short-term financial reports.The leading 5% of organizations are already realizing strong benefits, showing that success is possible but requires strategy, learning, and adaptation.
Lessons for Greece — and Especially for SMEs
For Small and Medium-Sized Enterprises (SMEs), the first lesson is caution and focus: don’t invest in AI simply because it’s trendy. Start with small, well-defined projects (e.g., automating customer support, text analysis, content generation) and measure tangible results.
Larger organizations, meanwhile, need to embed AI in strategic processes, not just run isolated pilots. That means investing in employee training, infrastructure adaptation, and long-term planning.
Only through this structured, strategic approach can Artificial Intelligence move from impressive pilot projects to real business impact.
See the report here:
Comments