AI Transformation Demands Business Overhaul, Not Just Integration, Experts Advise

Organizations must fundamentally transform their internal processes to truly harness the power of artificial intelligence, according to industry experts. Simply attaching AI to existing systems will not deliver the significant changes promised by AI providers. Denis Torii, vice president analyst at Gartner, stated, “It's a huge change and that's why the dynamic is not going to be as fast as people imagine.” He clarified that this is not solely an IT initiative or a push from business stakeholders, but rather an "organizational story, where the organization needs to adapt to it." Currently, many organizations are performing the same tasks with AI integrated, which Torii believes is unlikely to yield the expected transformation. He noted this on the sidelines of the recent Oracle AI World event in Las Vegas. A McKinsey report from August highlighted that organizations are deploying AI across enterprises using various tools, including chatbots, software coding agents, and autonomous systems. However, the enterprise-level financial impact remains limited, with only 37 percent of respondents reporting AI's contribution to their earnings before interest and tax (Ebit), a figure unchanged from the previous year, according to McKinsey's Global Survey on the State of AI.

Gabby Fredkin, head of analytics and insights at Adapt, emphasized that benefiting from AI requires a shift in leadership mindset. A fundamental change is necessary to move from merely layering technology to rethinking the organization's ethos, operations, and roles. Fredkin described it as a "complicated process." He added that successful AI adopters typically have leadership with a strong conceptual understanding of AI who foster innovation. The McKinsey report also pointed to a small group of high performers who are successfully utilizing AI. These organizations redesign workflows enabled by AI rather than inserting AI into existing ones, and they support their AI deployments with leadership commitment and operational rigor to achieve tangible gains.

Fredkin advised companies to examine their operating models to discover how they can transcend traditional processes and generate new revenue streams. He posed a hypothetical for a bank with extensive data from infrastructure investments: should it venture into infrastructure as an investment or establish a real estate investment trust? This strategic re-evaluation, he suggested, involves considering existing assets and operations and exploring parallels outside their current scope. He acknowledged that such a shift would present varying degrees of difficulty for different organizations.

The McKinsey report also indicated that AI-related operating costs are beginning to limit its use for approximately one in five organizations, despite most anticipating increased AI investments in the coming year. John Reed, co-founder of tech platform diginomica, noted that companies are adopting a cautious approach as they prepare their data and processes for AI integration. "If companies know that it's not going to be that expensive to use these technologies, I think they are going to find significant benefits," Reed stated. He believes AI could be particularly valuable for complex, mission-critical tasks that humans find challenging to execute in real time, such as assessing global supply chain impacts. However, he cautioned that adoption will take time due to challenges like fragmented data and legacy systems.

Investment in AI is rapidly increasing, with Gartner forecasting worldwide spending to reach $2.7 trillion this year, a nearly 50 percent annual increase. AI infrastructure, particularly data centers and specialized servers, is the primary spending driver, with Gartner calling the buildout "the largest infrastructure project humanity has ever undertaken." Goldman Sachs Research estimates global AI-related investment at $1 trillion this year, with $581 billion in the US, and projects AI capital expenditure to rise from 0.9 percent of global GDP in 2026 to 1.4 percent by 2028. However, this spending surge raises questions about the revenue required to justify it. A Bain report suggests the AI industry needs to generate $6 trillion in annual revenue by 2031 to validate the deployed capital.

"Return on investment is still one of the biggest issues [with AI]," Torii reiterated. "People are putting all this faith in AI, [with the assumption that] there's a lot of potential." He observed that "When you look at the broad picture, it's still lacking evidence that return on investment is there," though he acknowledged that vendors must invest heavily upfront and assist customers in their AI journey, overcoming significant friction and inertia. Torii compared AI's current phase to the early days of cloud computing, where substantial infrastructure investment preceded widespread adoption, questioning how long it will take for AI to bridge this gap.

Measuring returns presents another hurdle, according to Fredkin. "They [organisations] feel a productivity uplift. They feel better employee engagement. Putting an actual figure behind that is incredibly complex." He explained that for CFOs, ROI calculations have evolved, becoming less of a direct line item and more of a dynamic metric, citing the emerging term "double T Ebitda" (earnings before interest, tax, tokens, depreciation, and amortisation).

Unlike blockchain and metaverse technologies, AI is already gaining traction across businesses and is poised for wider adoption, Reed noted. Looking ahead, he expressed some pessimism regarding AI's broader implications, citing concerns about surveillance, disinformation, deep fakes, the erosion of trust in content, and job displacement. However, he also acknowledged the potential for technology to be decentralized and appropriated for positive uses by thoughtful individuals working on worthwhile projects.