One of the oft-cited benefits of generative AI is that it will be able to perform menial tasks and relieve us of workplace drudgery. For those at an executive level, that will most likely involve assisting with one task in particular: meetings.
According to a Harvard Business School study of 27 CEOs and 60,000 hours of work, 72% of chief executives’ time is spent in meetings. Relieving C-suiters from some of these meetings could allow more time to be spent on other business areas, such as shaping strategy, nurturing culture or coming up with new product ideas.
Some chief executives are already experimenting with this technology. Klarna CEO and co-founder Sebastian Siemiatkowski has sent an AI double to speak during a recent financial update. Sam Liang, the CEO of transcription software company Otter.ai, has developed a “Sam-bot” to eventually send to meetings in his stead. And Zoom CEO Eric Yuan used an AI avatar to deliver the initial comments in the company’s Q1 earnings webinar in May.
The video of the Zoom boss – which utilises a new feature called AI Companion – is slightly uncanny. But Yuan envisions a future where his digital twin can make decisions on his behalf and he wants to be in a position to send this replica to meetings within the next 12 months.
Steve Rafferty, head of APAC and EMEA for Zoom, is one of the executives who has experimented with the new technology. While the tool is not yet ready to interact with other people or answer questions, it has helped him deliver more personal messages to employees.
“My team stretches from the Arctic Circle all the way to Antarctica,” he says. “And there’s roughly 60 different languages spoken across those regions.” Hosting meetings can therefore be a challenge. But last month, Rafferty used his AI avatar to introduce a quarterly meeting in fluent French. “It means I can be in multiple places at once, speaking different languages,” he adds.
Rafferty claims his teams have “bought into” receiving messages from his AI body double and he describes the tool as “another string to the bow for business communication”.
Handing over power to AI
While this method of communication is currently one-way, Zoom is training its executives’ AI companions on their communication and decision-making style. “I always have my AI companion turned on,” Rafferty says. “It’s across my messages, video, phone and chat. It’s across everything.”
The next challenge, he adds, is trusting the AI to make decisions on his behalf.
Dan Thompson is CEO of Sensay, a startup that develops AI replicas of employees. Thompson says he is already using his own digital twin to make some decisions.
It’s another string to the bow for business communication
Sensay’s AI is trained on an individual’s calls, messages, emails and documents to learn how they speak and write and surmise their likely response to some questions. Thompson uses his own AI replica to draft replies to emails and messages, which he estimates saves him hours per day.
“It’s really useful for improving your efficiency,” he says. “I was sorting my visa at the embassy this morning, which took two hours. While I was there, all my morning emails had already been pre-drafted and all I had to do was read them and decide whether to hit send or edit.”
AI impersonators
While there are clear time-saving advantages to using AI in this way, it does come with a new set of risks.
AI remains prone to hallucinations, where the technology presents misleading or inaccurate information as if it were fact. This could be particularly detrimental if the false information is delivered under the guise of a company CEO.
Last year, the voice of Mark Read, the CEO of the multinational advertising company WPP, was imitated by fraudsters using deepfake technology. The scammers used the voice clone to try and convince a leader at one of WPP’s agencies to set up a new business, in order to solicit money and personal information.
If employees become used to hearing messages delivered by AI versions of their CEO, the trend could leave them more susceptible to falling for these types of scams.
You can’t just go blindly into this wild west of AI
Rafferty says it’s essential for businesses to develop a strategy for how they use this technology to communicate, which details what people can and cannot expect to see from someone’s AI companion.
“You can’t just go blindly into this wild west of AI,” he adds. “A lot of people are selling AI as the answer to everything, where it actually could be the problem. You need proper governance, structure and processes in place so people can trust it.”
Training these executive AI doubles also requires giving large language models access to sensitive company information. Rafferty, for example, has given his AI companion complete access to his work phone and recordings of his meetings.
Last year, one Microsoft employee warned that its Copilot AI programme could be used by employees to access sensitive HR documents or read executives’ emails, if the security is not set up properly.
These types of security issues are down to human error, rather than AI, according to Rafferty. “That’s down to the way the platform is set up,” he adds. “It can be told to only share data that’s relevant to people at a certain level in the business or make it available to everyone. If the planning is poor, you can expose yourself.”
Can AI replace the CEO?
For all its time-saving benefits, there are some elements of the CEO role that will most likely remain irreplaceable, no matter how advanced AI replicas become. “The CEO role isn’t just about outputs – it’s about meaning. AI can’t embody organisational purpose, grapple with ethical dilemmas or inspire through shared struggle,” says Dr Alexandra Dobra-Kiel, director of innovation and strategy at the consultancy Behave.
Certain business decisions will always need to be made by a human and a bot cannot mimic the leadership, direction and inspiration a CEO can provide. “AI can analyse trends, but it can’t navigate the ambiguities of long-term vision, trade-offs or the instinctive ‘gut calls’ that define high-stakes leadership,” Dobra-Kiel adds.
Although Yuan and other tech CEOs share the ambition of offloading some of their heavy meeting schedules to their AI assistants, there remains a human element of leadership that these AI imitations are unlikely to replace. “People follow leaders they believe in, not just voices that sound like them,” says Dobra-Kiel.
One of the oft-cited benefits of generative AI is that it will be able to perform menial tasks and relieve us of workplace drudgery. For those at an executive level, that will most likely involve assisting with one task in particular: meetings.
According to a Harvard Business School study of 27 CEOs and 60,000 hours of work, 72% of chief executives’ time is spent in meetings. Relieving C-suiters from some of these meetings could allow more time to be spent on other business areas, such as shaping strategy, nurturing culture or coming up with new product ideas.
Some chief executives are already experimenting with this technology. Klarna CEO and co-founder Sebastian Siemiatkowski has sent an AI double to speak during a recent financial update. Sam Liang, the CEO of transcription software company Otter.ai, has developed a “Sam-bot” to eventually send to meetings in his stead. And Zoom CEO Eric Yuan used an AI avatar to deliver the initial comments in the company’s Q1 earnings webinar in May.
