The business world has identified generative artificial intelligence (GenAI) as a strategic priority, and billions of dollars are being invested in this area (in developed countries, of course). However, the newly published "The State of Artificial Intelligence in Business 2025" report, prepared in collaboration with MIT and Project NANDA, reveals that these investments are struggling to deliver the expected returns.
According to the report, 95% of organizations are failing to achieve measurable returns from their initiatives in this area. This situation is referred to in the report as the "GenAI Gap." On one side, there is a minority of 5% who have successfully integrated AI and generated tangible value, and on the other, there is a majority of 95% who have failed to advance beyond the pilot project stage.
This striking picture, rather than a failure, points to a critical crossroads that emphasizes the importance of the right strategy, and experts point to the following strategies for leaders who want to be on the right track:
1. Core Problem: Learning Insufficiency
The report's most important finding is that the primary reason behind the failure of pilot projects is not a lack of technology, but rather the systems' inability to "learn." Most enterprise AI solutions remain static tools that fail to learn from feedback, adapt to workflows, and improve over time. This leads employees to perceive these tools as ineffective and not use them. Conversely, successful organizations prefer dynamic systems that continuously learn, understand context, and evolve with business processes. To overcome organizational design and learning insufficiency, companies should focus on fundamentally redesigning their workflows and organizations rather than incorporating GenAI into their existing processes and structures.
2. Shadow AI: An opportunity, not a threat
When enterprise tools prove inadequate, employees tend to use personal AI tools (ChatGPT, Claude, etc.) for business purposes. While only 40% of organizations reported purchasing a formal Large Language Model (LLM) subscription, employees at over 90% of surveyed companies regularly use personal AI tools for their work tasks. This phenomenon, which the report describes as the "Shadow AI Economy," is actually a valuable source of insight for leaders. Analyzing which tools employees prefer and for what purposes provides insight into the organization's true needs and the features required for successful AI integration.
3 Investment Paradox: The Hidden Treasure in the Back Office
While approximately 70% of AI budgets are allocated to sales and marketing functions, back-office automation typically delivers a better return on investment (ROI). According to the report, the real gains come from eliminating business process outsourcers and external agencies, rather than from internal staff cuts. For example, eliminating outsourcing in areas like customer service and document processing can save $2 million to $10 million annually, while external creative and content costs can be reduced by 30%. A large portion of AI budgets is directed to front-office functions like sales and marketing, where return on investment (ROI) can be easily measured. However, the report shows that the highest and most sustainable productivity gains are often hidden in back-office operations like operations, finance, and supply chain. Successful companies achieve significant savings by reducing outsourcing and agency costs rather than reducing staff.
4 Strategic Roadmaps: Partner, Don't Build
The report's clearest advice to leaders is to avoid building AI solutions in-house. Building strategic partnerships and adopting specialized, learning-capable, off-the-shelf solutions has a success rate twice as high as in-house development efforts. Of course, the right partner is not just a technology provider; it's a companion in achieving business goals.
In conclusion, the GenAI Chasm is not permanent, but crossing it requires making fundamentally different choices about technology, partnerships, and organizational design.
