AI transformation has barely reached the core operations of large companies and governments, according to Marc Warner, Accenture’s global chief technology officer, who said organisations must focus less on isolated tools and more on redesigning how work is done.
The number of large enterprises or government processes “that have been really, truly reinvented with AI at the absolute core is basically close to zero”, Mr Warner said during a virtual Fortune Brainstorm AI event.
He said many businesses had concentrated on peripheral uses such as marketing, meeting notes and transcription. Coding was a partial exception because its returns could be measured more easily, he added.
AI transformation is an organisational challenge
Mr Warner said the biggest economic gains would come from applying artificial intelligence to core business processes, although doing so was considerably riskier. He compared the current stage with the early use of electricity in factories, when it was used for lighting before production systems were redesigned around electric motors.
Greg Ulrich, Mastercard’s chief technology officer, said companies would achieve greater returns by rebuilding entire workflows rather than adding AI to individual tasks. Mastercard has been redesigning its product development process with smaller teams and faster turnaround, he said.
But accelerating one stage of a long process can simply create a backlog at the next stage. “The AI agent to automate a piece of this isn’t… the long pole in the tent for a lot of this,” Mr Ulrich said. “It’s the rethinking. It’s the remapping.”
Mr Warner said he had previously viewed AI programmes principally as technology projects, but now regarded the wider transformation as a “politics project”. Matthieu Caillat of AXA said the challenge was increasingly about bringing employees and other stakeholders along as the technology developed.
Mr Ulrich said Mastercard relied on support from its chief executive, including weekly updates with the chief executive’s direct reports. However, he added that no organisation had yet solved the political challenges of large companies.
Executives urge experimentation as technology changes
The panel also considered who should be held responsible when an AI agent fails. Mr Warner said accountability would usually be divided between the business owner, who was responsible for the agent’s value, the technologist, who was responsible for its operation, and the chief information security officer, who was responsible for security.
He cautioned against allowing concerns about blame to slow experimentation, except in cases involving critical incidents such as a major security breach. “I don’t think most organizations should be worrying too much about the blame game right now,” he said, arguing that employees should receive credit for trying ideas that might not succeed.
Marjorie Janiewicz of Mistral AI, who said she had spent a decade in cybersecurity, agreed that finger-pointing was unhelpful. She said companies should instead ask whether they controlled the technology they adopted and could establish their own safeguards.
The pace of development is also forcing companies to revise projects quickly. Mr Ulrich said tools Mastercard had built about a year earlier were now “quaint, and just not that useful today”, while Mr Caillat said AXA could no longer afford to “measure twice and cut once” because conclusions could become outdated before they were certain.
AXA initially identified about 400 AI use cases to build enthusiasm before narrowing its focus. Mr Caillat said its plan for 2027 to 2029 aimed to generate between €500 million and €700 million a year in recurring pre-tax value from AI by 2029, while the insurer was beginning to use about four billion unstructured documents.
Mr Warner said Accenture expected the pace of development to continue until something approaching 2030. He said energy constraints in Western countries could begin to have an impact in the early 2030s.
