Getty Images
CEOs see business case for creating AI-powered chief execs
CEOs consider AI virtual counterparts to handle routine tasks and provide feedback, boosting efficiency and supporting strategic decisions.
There's a way to create a virtual CEO using AI. First, build a database using all the emails, reports, studies, presentations and video meetings that a CEO has seen, penned or voiced. From there, the AI can glean a CEO's problem-solving approaches, priorities and methods for examining proposals.
Managers could use this AI CEO to test ideas, get feedback on a proposal and mine it for specific company issues. The real CEO might rely on the virtual CEO to handle some routine matters. Once the CEO retires or moves to a different job, the chatbot could continue to offer advice and insights. That might be valuable, according to some CEOs.
Another question is whether businesses should do this. Research is emerging on the ethics of creating human digital twins depending on how closely they resemble someone, such as replicating physical appearance and mental states. Digital twins are a fraught area. The very idea of using AI for this purpose is raising alarms about its potential use for manipulation. Even if a person grants permission, a digital replica could raise questions about personhood or what it means to be human.
Nonetheless, some are trying out the concept. Roger Hamilton, CEO of Genius Group, an educational technology and education company, built a chatbot based on Alan Turing, the famed computer scientist whose 1936 paper on computation described a machine capable of performing calculations. In the 1950s, he devised what became known as the Turing Test, or a means of determining whether a computer is thinking like a human. The chatbot was built using OpenAI's GPT-3 language model as its core, with additional capabilities such as memory and real-time internet access added on top. It is enabled through an integration with Microsoft Copilot.
Hamilton is convinced this AI version of Turing is applicable to current business issues. In April, it appointed its virtual representation of Turing, which also uses the real Turing's likeness, as the company's chief AI officer.
This AI version of Turing wrote a white paper outlining its version of the future of AI and the role of education within it. Hamilton said the company was impressed with the work, and the AI rose to the top of its chief AI officer recruitment list. "It already knows more than any other candidates we're finding for this role," he said.
Sam Glassenberg, CEO of Level Ex, a company that develops video games designed to train doctors, is sceptical about the value of a chatbot based on Turing.
Chatbot can't compete
Turing "came up with some of the fundamental theories of computation out of whole cloth," he said. "I highly doubt -- at least based on the technologies of today -- that the chatbot is going to come up with new theories and insights on its own."
Sam Glassenberg
But Glassenberg does see value in creating an AI CEO -- something that could take the chief executive's data record, emails, conversations and organizational knowledge and then provide context and an explanation of what happened on various issues to the existing CEO, others in the firm and a successor.
Just providing a "synthesis" of a CEO "is valuable to a new CEO, to employees -- if you can do it," Glassenberg said.
Similarly, the AI CEO could be used as a "sparring partner" -- a way for others in a company to check a perspective, said Rob Whiteley, CEO of Coder Technologies, which provides a cloud development environment for businesses. That capability might allow him to use his time more efficiently. "I don't have to be in every meeting or in every interaction."
If a CEO decides to build a digital replica, there are concerns about whether it will be useful, especially as time passes.
Pedro Amorim
The AI CEO needs enough information to make it work, said Pedro Amorim, a professor of industrial engineering and management at the University of Porto in Portugal as well as co-founder of analytics consultancy LTPlabs.
A CEO chatbot is potentially powerful, but "how much data can we get on a singular level, on a personal level, to make it fly?" Amorim said.
Amorim believes an AI CEO that combines the information of multiple CEOs might be less myopic and more effective in providing a broader, more in-depth and comprehensive analysis of a business problem.
Building method
A method for building this AI CEO that might work best is taking specific information from CEO or decision-maker documents and feeding that into a large language model. This approach gives the company control over how much and what type of data to include. However, it's also crucial that these AI CEO models relate to current conditions, which might be the biggest problem facing their usefulness in adding business value.
Whiteley and Glassenberg see a limit to an AI CEO's ability to solve high-level problems and adapt to a changing world.
Rob Whiteley
For instance, it might be helpful for a new CEO to access a chatbot during a period of transition. That AI CEO might work "for some matter of months before it begins to be unadaptable to current business conditions," Whiteley said.
Amorim questions the long-term usefulness as well. "Will the model be able to relate the past to the real world?"
Turing died in 1954, and Hamilton said the Turing chatbot has delivered significant assistance to his company, including helping it devise a five-year business strategy and ranking system for one of its projects. Without the chatbot's help, "it would have taken us much longer to pull it all together," he said.
The Turing chatbot is "now in every executive team meeting," he said.
Patrick Thibodeau is an editor at large for TechTarget Editorial who covers HCM and ERP technologies. He's worked for more than two decades as an enterprise IT reporter.
