Getting AI to train your spokespeople
By Jeremy Chan, Capital Markets Content Strategist, Aspectus
Media training is more than just testing for a sound bite
AI is being worked into everything. From advertising to big tech to finance, every industry is using it with the understandable goal of taking on mundane and manual work to free up employees to focus on more high-value tasks.
As AI becomes more embedded in the way we work, there is a risk it gets used in tasks where its value-add is questionable.
Media training has traditionally been the purview of ex-journalists who have made their way to the dark side. Who better to explain the media interview process than someone who has sat in the newsroom and has done thousands of them?
Recently, some PR firms have started offering an AI trainer that can coach company execs about the dos and don’ts of interviewing. Some offerings are quite tantalising in their pitch. A mock interview could produce in near real-time a mock article, written in the style of the reporters a firm is targeting.
That is an attractive prospect. A spokesperson can see how their interview will translate into results. They could test and refine their answers to create an ideal response for every media opportunity.
But speaking from the perspective of someone who been on the other side, this is an incomplete understanding of the news-gathering process:
For one, you’re almost certainly not the only voice. A journalist will speak to many sources when building a story. There is no guarantee what you say is how a journalist will angle their article, or even if you’ll be included.
Every spokesperson wants to be quoted, and every journalist wants a nice juicy sound bite, but that’s 10 seconds in what are typically 20-30 minute interviews. The majority of an interview is fact-finding, where a journalist will ask, question and probe to give readers or viewers the context of why their covering a story.
News is an inherently relationship-based affair. A source who trusts you may give you a tip and a PR with whom you have good rapport may give you an exclusive. The ideal scenario for any journalist (to the dismay of the comms team) is to have the direct cell number of any source.
Media training is not just honing your message. It involves understanding how the sausage gets made and prepping spokespeople on how to interact with journalists.
One investment bank that hosts an annual off-the-record gathering between media and senior staff required all bankers attending to undergo media training, lest they accidentally give a tip to a member of the press.
AI’s role in comms and media
There are real benefits to AI, but realising them requires a firm understanding of its limitations. As a journalist, I used it extensively to transcribe interviews: a major time saver when I had hours of audio tied to longform features. An AI transcription of a mock interview can likewise help in rapidly reviewing responses during a media training session.
Use of AI often runs into trouble when the end user – be it a reader or client – sees its results without human accountability.
Earlier this year, Apple shelved its AI news alerts after it gave out inaccurate headlines. A lack of due diligence from both the journalist and the editor’s desk allowed a summer reading list that was syndicated in several US papers to include books that did not exist.
I’ve spoken with executives at some of the world’s largest banks and financial firms and many highlighted how their company was incorporating AI into employee workflows. It is creating a first draft for client proposals, code co-piloting for programmers and summarising documents to generate memos for meetings.
All were equally reluctant to push it to client-facing functions, and many needed to fully understand the data a model was trained on. The potential for hallucinations or misinformation was not worth the reputational risk, especially in an industry where trust is of the utmost importance.
As a journalist you are paid not only for your words but to take responsibility that what you write is accurate, true and fair. It is your byline on an article. The same can be said for media training.
AI excels at certain tasks but we’re at risk of anthropomorphising it if we allow it to teach. Most of the AI currently in use is in the form of large language models. These do not think or understand what it is saying; only predicting the most probable next word.
For something as important as getting executives in front of national or international press, firms looking to train their spokespeople are placing a great deal of trust in technology designed to mimic (not recreate) human interaction.
The crucial question must be: Why get a machine to teach you the best way to speak to a human?
Key takeaways
Can AI replace human-led media training?
No. It assists with prep but can’t replicate newsroom insight or relationship skills.
Where does AI add real value?
Efficient transcription and quick draft responses during mock interviews.
What are the risks of AI‑only training?
Hallucinations, lack of accountability, missing nuance and context.
