When Lech Mintoot Cheese wrote speeches for Ben van Beurden, part of his job was to help the then Shell CEO connect emotionally with audiences in a way he was not accustomed to.
The oil and gas company has been under pressure from investors and activists for its role in enabling climate change, and whenever Van Beurden spoke, it was “a very impressive series of precise decisions,” Mintout-Ces says. It took a long time to build a relationship with Van Beurden and gain his trust.
Speechwriters like Mintowt-Czyz now face a new set of pressures as generative AI tools like ChatGPT and GPT-4 promise to create persuasive speeches based on simple prompts for topic and tone and the introduction of a few key quotes. Other tools have emerged to aid delivery, such as analyzing audio recordings or live speeches to detect emotions and evaluate their impact.
As business leaders' opinions come under more scrutiny, especially from younger employees, communications — from employee update emails on geopolitical events to investor presentations — are taking on greater importance. An incorrect quote can easily lead to backlash from employees or customers.
AI tools can help produce speeches quickly — producing thousands of words in a matter of minutes — and help executives seeking to engage with employees or stakeholders more effectively. But there are obvious downsides: their content may lack personality and originality and may be full of errors.
“It's a lively discussion in the speechwriting community,” Mintoute-Sizes says. Until now, AI tools “don’t understand what words mean. Just how they relate to each other and how the words are put together. He's very good at it, but he doesn't fundamentally understand what he's writing, and what effect it has on people. . . At this stage, it is not possible to distinguish between good, bad and average.
While Mintot-Chese has “certainly been playing around with the AI tools to learn about their capabilities,” he says he hasn't used them to help him write anything professionally.
Until now, communications consultants and corporate speechwriters tend to use technology as a kind of digital assistant, rather than taking its content literally. It's just one aspect of the new AI-powered leadership style taking shape across businesses: other tools are available to help CEOs with tasks like scheduling, time management, and acting as a pressure-testing partner on key decisions.
Artificial intelligence can generate data to inform and enrich the content of speeches or presentations. Mintoot Chase says he might use it “as a kind of semi-intelligent lexicon.” Other communications professionals use tools to paraphrase speeches — something they can do in a fraction of the time it would take a person — to delete clichés or make the speaker seem more energetic or positive.
For example, PR company Kekst CNC has worked with the University of Oxford on technology that enables its clients to see at a glance their linguistic tendencies and suggest improvements. It provides an “impact score” based on how an executive uses pauses, pacing, emotional language and storytelling, among other metrics, after analyzing audio recordings — from earnings calls and internal meetings to media interviews. Data from an executive can be compared to a broader group to see how that individual is performing compared to his or her peers.
Giving speeches is just as important as telling, says Shelagh Ball, chief corporate affairs officer at Canada Retirement Fund OMERS. “Delivery is important. That's what makes the message memorable,” she says. “A data-driven perspective can provide an objective check against bias. We will never rely on the tool alone to provide advice. But while nuances in delivery can have a beneficial impact, I will leverage our expertise with resources that can give us a competitive advantage.
Cheryl Clark, a speechwriter and executive ghostwriter, works with ChatGPT as a junior writing assistant, tasking her with summarizing longer documents, engaging in collaborative idea generation and general research.
“A healthcare speechwriter might write a keynote about emerging technologies,” says Clark. “Using ChatGPT, they can quickly gather and summarize recent developments and relevant statistics. AI's ability to sift through massive amounts of data can save hours of research time… and help a writer discover key trends they've missed.”
Clark notes that she wants to “harness the speed and processing power” of an AI tool for rudimentary tasks, but will continue to create the “emotional, compelling, and original story” herself because a robot cannot yet “feel and experience the way we do.”
She adds: “People who depend on it [AI tools] To replace human talent will quickly find its output generic. . . Tools like ChatGPT are just that, tools. It does not replace human creativity.
Clark uses PlayHT, an AI company that builds conversational audio models capable of replicating any voice or accent and generating speech in real time. The audio tool can help anyone who “needs professional audio but doesn't have a big budget to go into the studio with a human.”
Most advisors to senior executives are wary of using AI tools in sensitive communications. They say that if employees discovered that a highly emotional ad — such as a note to employees about being laid off — had been generated by a robot, it would convey a lack of respect and empathy.
AI tools can be used to get feedback on an important decision, says Euro Benat, global head of AI and data science at technology investment firm Prosos. “You might think, ‘I haven't thought about that,'” he says, but he warns, “I would be very careful not to use these tools for anything physical to the company.” He adds, “I will always have control and will never automate anything.”
Tera Alas, director of research and economics at consultancy McKinsey in the UK and Ireland, agrees that anyone using these tools today needs a human “in the know” to ensure there are no errors or biases. “In terms of language, there are a lot of things that AI gets wrong,” she says.
Executives also need to realize that AI-generated content is discoverable. Technology is being rapidly developed to detect AI outputs.
Then there is the point of authenticity to consider. “If you think forward in a world where all communications are generated by AI, AI will read them too,” Alas adds. “Employees may not read their email. There is a loss of meaning unless you put it back into a real-life context.
Mintout-Sez agrees: “What is communication? It's about human connection. One human reaching out to another. Either to inform or to persuade or whatever. . . . When you delegate an element of that to a computer, it's no longer human-to-human communication.” .
For now, he is betting on himself. He adds that his work is “more than just words on a page.” “People should trust those words. How can you trust that AI knows what it's doing?
