Accenture's Paul Daugherty: Generative AI today, but watch what's next

By Larry Dignan

Generative AI will have a tremendous impact on workers and likely impact 40 of the working hours across industries, but that doesn't mean 40% of jobs will go away, said Accenture CTO Paul Daugherty, who was the 1,000th guest on Constellation Research's DisrupTV.

Daugherty also said on the show that generative AI is just the first installment of what's likely to be a set of innovations that'll change business. "As exciting as generative AI is, it's not the last and probably not the biggest breakthrough we'll see in AI. There will be other bigger breakthroughs coming with common sense AI," said Daugherty.

 

He added that shared reality and the merger of the physical world and digital bits will be another advance. "The next stage of digital is digital plus physical," said Daugherty, who also said metaverse has potential even though it's being panned on multiple fronts.

Other technologies to watch include:

• Computational chemistry.
• Synthetic biology.
• Generative AI in pharmaceutical and drug discovery.
• Operational technology with new forms of computing like quantum computing.

Daugherty said his team at Accenture is already working on its next Tech Vision Report. In the meantime, here are some generative AI takeaways from Daugherty.

• Generative AI will create automation that can replace some jobs, but the bigger impact will be human augmentation. "Every worker will have a co-pilot or multiple co-pilots that help us do things more effectively," said Daugherty. "AI will give people superpowers in the form of these co-pilots that could allow us to do more things."
• Think of generative AI as a bit of Moore's Law for people where it can help us know more and extend skills.
• Enterprises need to step back and get perspective on generative AI. For instance, companies can't assume generative AI will solve everything. Instead, businesses need to look holistically and see where to apply the technology.
• Figure out what models you want to consume and then tune. "For enterprises there is going to be an array of models to just consume or in some cases fine tune and customize," explained Daugherty. "In other cases, you'll want to develop your own models for unique domains."
• Companies will need to change processes. "Processes are so different with co-pilots," said Daugherty. "There's also the change management and training around it."