Many companies have AI strategies, but few can prove their effectiveness, according to a report from KPMG, one of the world's largest audit firms.
Companies are rushing to adopt artificial intelligence (AI), but many still struggle to demonstrate its effectiveness, Matthieu Wallich-Petit, Director of Customer and Market Operations at KPMG France, told Euronews Next at VivaTech in Paris.
According to the company's report published in March, 95% of its clients already have an AI strategy in place, and 64% have already obtained concrete results from the technology. But only 8% can clearly measure the return on investment.
"Our clients are indeed building serious AI strategies, but in practice, on the ground, the gap is still very large," said Wallich-Petit.
According to him, KPMG's task is to help companies close this gap, as technology develops faster than most organizations can adapt. "It's striking that the pace of technological acceleration is truly exponential," he noted. "And within companies, we see adoption that remains almost linear."
According to the survey, only about 10% of KPMG's clients have already implemented AI company-wide.
In the insurance sector, he noted, companies are beginning to apply AI not only for claims settlement. "If before it was mainly about automating the handling of claims, now it covers the entire cycle—from scoring new clients and pricing to servicing," he said.
According to KPMG, companies continue to increase AI budgets, as boards view these technologies as a competitive advantage and a way to attract talent. At the same time, businesses are increasingly monitoring whether such investments provide a fast and transparent return, Wallich-Petit noted.
Advice for companies
The French top manager also emphasized that when transitioning to AI use, the focus for leaders should be on employees. "In my opinion, it all comes down to people, not technology," he believes. "Upskilling, training staff, is probably the most important strategic success factor for any AI strategy."
For companies stuck between pilot projects and large-scale deployment, embedding AI into daily business processes should be a priority, Wallich-Petit added. "A recipe for success is to move from proof of concept, from piloting, to actually embedding AI into processes," he said.
According to Wallich-Petit, this also involves stricter governance, better data management, and additional employee training.
He also stressed that AI sovereignty is becoming an increasingly important issue for companies, especially given businesses' dependence on a limited number of powerful model providers. "The main idea is not to rely on a single model, but to have a diversity of models," he said.
This issue has become much more concrete as access to a number of advanced AI models is increasingly tied to geopolitics. In May, KPMG and the American AI company Anthropic announced a global alliance: Claude will be integrated into KPMG's platform for working with clients, and the company's employees worldwide will gain access to this AI assistant. A few weeks later, Anthropic reported that, by order of the US government, it must suspend access for foreign citizens to its Fable 5 and Mythos 5 models.
More on this story in the video at the top of the page.
