Provokeglobal:“Thought Leadership Requires An Original Thought”

(MENAFN- PRovoke) CHICAGO - AI-generated thought leadership risks eroding trust and stripping brands of their distinctive voices, communications leaders warned at PRovoke Media's Global Summit, urging agencies to protect original thinking as they embrace the technology.

Panellists said pressure to produce content faster must not come at the expense of the opinions, experience and judgment that make communications valuable, with Finn Partners senior partner Alev de Costa arguing that credible thought leadership still required an increasingly important ingredient: a thought.

Spark Public Relations senior vice president Susanna Kalnes said she avoided giving C-suite clients questions before thought leadership interviews because they might use AI to prepare their answers, leaving communicators with a version of everybody else's thinking rather than a distinctive perspective.

“AI will give you the thought of the masses. A thought leader will give you their thought, and that's the difference,” she said.“In order for thought leadership to work, you need to know that. I work with a lot of C-suite and I ask them questions for thought leadership. I don't give them questions in advance. You know why? Because if I give them the questions in advance, they might prep with AI, and I don't want that. I want their unique point of view, and that's the only way it works now.”

The observation captured a central concern of the session, 'Real Stories, Artificial Intelligence: Governance, Transparency and Trust in the AI Era': how an industry increasingly using AI to produce work more efficiently can preserve the originality, credibility and human judgment on which its value depends.

Moderated by Davis+Gilbert partner Samantha Rothaus, the discussion brought together Kalnes, de Costa and Gabby Schiavenato, founder of TinyGoat, to explore the decisions behind responsible AI adoption, from what agencies tell clients to how they develop junior colleagues.

De Costa said the proliferation of AI content was creating renewed appetite for an identifiable human perspective. While the technology could support efficiency, the substance of thought leadership still had to come from the individual.

“I've noticed very much a return to a desire for the real thing, which I think is great,” she said.“For thought leadership, you have to have a thought. You have to have an opinion. You have to have the knowledge that you brought over your career to back that up. So I don't think there's a replacement for that.”

The cost of 'getting away with' AI

Schiavenato drew a distinction between using AI for repetitive correspondence, such as a quick sales follow-up, and allowing it to stand in for a person's thinking. It could help someone explore a subject when they were stuck, but that required a different relationship with the technology from simply accepting its answer.

“I firmly believe that we probably shouldn't have a lot of thought leadership used with AI unless you are hoping that maybe it'll spark a thought for you,” she said.“You're prompting it like, 'Hey, help me out, I'm in a little bit of a rut when it comes to a certain topic and I want to think about it.' But it's dangerous if we start to use it as something that is the masses and not an actual person's thoughts.”

For Rothaus, the implications extended beyond the quality of the content to the relationship with its audience.“If the audience believes in what they're reading and then finds out it was written by an AI, that can feel deceptive,” she said, adding that helping clients avoid misleading practices was closely connected to communicators' responsibility to build trusted voices.

Schiavenato suggested that greater openness could also force organisations to confront whether they were comfortable with their own use of the technology. If disclosing AI involvement felt embarrassing, that was a reason to reconsider how it was being used.

“We lose our uniqueness. Each corporation, each organisation loses their uniqueness if we start to just be AI or AI obvious,” she said.“I think we're in this weird space right now where we think we're getting away with using AI and it's not necessarily true.”

Where agencies draw the line

That did not mean abandoning the efficiencies AI offered. Kalnes, whose agency represents AI businesses and other emerging technology companies, said some clients actively expected their communications partners to use it.

“I rep a lot of AI companies and I think the expectation is we're using AI in every way that we can to be efficient for the business, because if we're not, we're wasting their money,” she said.

The more useful question was where those efficiencies belonged, and where a human needed to remain responsible for the work. Kalnes said reporting, gathering news to inform potential newsjacking opportunities and iterating plans or press releases were all areas where AI could reduce time spent on tasks while leaving practitioners to make the important decisions.

“I'm the biggest supporter of AI when it's used to do the things that otherwise take too long and we bill too much for,” she said.“We're still having that judgment as to what we pitch and when, but can we get those things to us faster? Can we iterate a press release faster? Can we iterate a plan on a page faster? Sure. And that's where we can pick up efficiencies.”

She had begun making those distinctions explicit in pitches for new business, explaining to prospective clients both where AI would be used and where the agency would draw the line.

“I started putting in new business proposals slides that say where we use AI and where we don't use AI, because there's certain things that we don't want AI doing,” she said.“We don't want it making the decision in a crisis. We don't want AI writing the quotes on behalf of a CEO of a public company. So I think it's really important to call that out and be transparent with our clients.”

Buying a tool is not a strategy

The conversation also challenged the assumption that buying an approved AI product meant an organisation had brought its employees' use of the technology under control.

Rothaus raised the risks associated with“shadow AI”, where people use tools at their own discretion without the organisation understanding what information is being entered or how the outputs are being used. Schiavenato said companies could mistake access to a tool for an effective adoption strategy, while staff quietly continued using alternatives on personal devices.

“A lot of times I find companies that assume they've fixed it by turning on Copilot in their Microsoft suite, and then quickly their employees realise that's not very good AI, so I'm going to use my Claude on the side, I'm going to use my Gemini, whatever it is that I feel comfortable with,” she said.

Her starting point was to establish what employees were actually using, rather than what their employer assumed they were using. That could reveal different preferences across teams and help inform a more practical approach to implementation.

“It doesn't have to be a blanket LLM for the entire organisation. We can also specialise in different LLMs for certain parts,” she said.“And we need to have a policy for all of that and ensure that we're responsibly implementing and training folks.”

Crucially, Schiavenato argued that organisations needed to agree the policy before purchasing tools or training employees. She described a client preparing to undertake training while its policy was still being developed.

“To me, that's backwards,” she said.“We need to know what we as an organisation plan to have and want when it comes to AI, where we're going to have it, where we're not going to have it, before we're training folks on it.”

Rothaus added that a policy's existence was insufficient if employees could not put it into practice:“Having a policy is only as good as the people understanding what the policy is and how to actually implement it.”

The coffee shop experiment

The question of what practitioners understood, and how they learned it, became particularly pressing when the discussion turned to younger employees. As Rothaus pointed out, the judgment that senior people brought to AI outputs had often been developed through years of experience, trial and error and mentorship.

Kalnes said senior practitioners had a responsibility to teach colleagues how to assess what the technology produced, especially if those colleagues had used AI throughout their working lives.

“As the human in the loop on all of this stuff, you have to make sure you're looking at every output with a critical eye, and that's on you,” she said.“Right now, as senior practitioners, that really very much falls on us because we understand from years of being in the business how to discern. We also need to teach the younger generation how to discern when they've been maybe AI-native their entire career, and that is a challenge.”

Her response had been to pair AI training with exercises designed to develop skills without it. One, the“coffee shop experiment”, asked employees at every level to spend two hours away from Slack and email, observing their surroundings and writing 500 words.

“You don't use AI to write, you just write. And I want you to write with all of your senses,” she said.“What do you see? What do you smell? What do you taste? How does that make you feel? Write as though you are a novel writer, you are writing your very first novel, and come back with 500 words.”

The team then read their work aloud in a session Kalnes compared to a poetry slam. Beyond being enjoyable, the exercise revealed abilities that were not necessarily apparent in day-to-day agency work.

“What I realised in that process was what great writers I have on my team, and you don't always know it if what you're getting back is AI-informed from the first draft,” she said.

“When you are working with journalists, who live and breathe words, words matter. You have an entire dictionary at your disposal, and LLMs are just going to pull in the same words because that's what the masses are using, the same sentence structures. But what if you came to the table with some coffee shop type writing and what would that do?”

Afterwards, she said, the team's pitches began to improve. The exercise also spoke to a broader problem in media relations: approaches that once felt distinctive were becoming recognisable templates as AI reproduced them at scale.

“If I were a journalist and I read another pitch email that started with a statement, a colon, and then a provocative question, I might just throw my laptop out the window,” Kalnes said.“And that used to be a great way to start a pitch email.”

A second training exercise focused on listening and interviewing. Kalnes and her direct reports took turns recording five-minute podcast-style conversations on subjects of their choosing, from a puppy to women's basketball, to practise following an answer and asking the next question.

“The point behind all of that is to develop listening skills and to ask follow-up questions, because ultimately what you're doing when you're in a thought leadership meeting is you're running a podcast,” she said.

Bring the proof

Those conversations were also essential when communicating complex technology. Kalnes described a three-stage approach to working with AI companies: understand the product well enough to explain it to a child, establish why the intended audience should care, and provide evidence that supports the claims.

That meant talking to engineers, testing analogies with them and looking beyond an impressive description of what a technology could do.

“I get a lot of companies that say they're doing really amazing things. I get a lot of companies that have really awesome technology, but without the white paper to back it up, the outside sources validating it, the customer actually having tangible results, journalists will say, 'Who cares?'” she said.

“Sure, I'll use AI to speed up maybe putting a press release on paper. But that process cannot be done without human interaction.”

De Costa said the same responsibility applied to how brands appeared in AI-driven search. Organisations needed to check whether the information being surfaced was accurate and whether it needed correcting, while continuing to invest in credible material.

“You're responsible for what's put out there,” she said.“I still think we need to go back to the basics of putting the real thought leadership out there and putting the authenticity out there.”

Closing the discussion, Schiavenato urged organisations to assess their use of AI regularly, including identifying places where it should be removed as well as new opportunities to introduce it.

“I think sometimes we can eventually get lazy and go to it in a way of a quick fix instead of pushing ourselves to think more and to actually create,” she said.“Keep learning about it, because every single day we're behind.”

For Kalnes, the industry's task was to keep returning to the relationship at the centre of its work, however much the tools changed.

“AI is going to disrupt public relations a lot over the next couple of years, but at the core of what we do is we relate to publics,” she said.“As technology continues to get better and continues to help us, we always need to be circling back to: with this new iteration of what technology is doing, how are we as humans relating to publics?”

De Costa's final recommendation placed responsibility firmly with the practitioner:“AI is autonomous, but your use of it isn't. So you need to be a responsible implementer of the technology.”

MENAFN27102026000219011063ID1111874371