The Truth Algorithm: Sustainable Brands in the Age of Large Language Models
Recently I engaged in a discussion on Large Language Models (LLMs) and their impact on the global communications profession. The conversation was with current and former Chief Communications Officers through the Arthur Page Society, where I am a member. As communicators in business, we are all grappling with the best way to use AI to manage a company’s communications and overall brand.
As I was listening to the presentation on the state-of-play with LLMs, and how we as a profession can use AI to communicate and build brands more effectively, and in our case at DevryBV, sustainable brands, one question kept popping up in my mind: how can we ensure the truth is told about companies, their products, actions, and impact on people and nature on LLMs?
The truth. It’s a core value of our company. We place truth at the center of our Integrated Sustainability Framework and walk clients through our Seek>Face>Follow>Share the Truth model. Our logo is an open sphere representing the dynamic and continuous process leaders must engage in to bring truth into their companies and share truth with the outside world. In this “post-truth” world where consumers seek more information on companies and the products, services, capital, and technology they deploy in the marketplace, it becomes difficult to figure out the actual truth behind a brand, especially when information changes quickly.
How do companies manage their brand when they do not control the message or how and where consumers get their information? This is not a new business issue by any means, but managing brands is much more complicated in the age of AI.
Here is a lesson in truth from one of the preeminent truth tellers of the 20th century:
“I would like to say to the diligent reader of my writings and to others who are interested in them that I am not at all concerned with appearing to be inconsistent. In my search after Truth I have discarded many ideas and learnt many new things. Old as I am in age, I have not feeling that I have ceased to grow inwardly or that my growth will stop at the dissolution of the flesh. What I am concerned with is my readiness to obey the call to Truth…, and therefore, when anybody finds any inconsistency between any two writings on mine, if he still has faith in my sanity, he would do well to choose the later of the two on the same subject.”
M.K. Gandhi, TO THE READER, “India of My Dreams”
Taking a lesson from Gandhi’s message on truth in leadership, how might business leaders live and share the truth of their business and its brand?
Handing the keys to a company’s brand to Artificial Intelligence (AI) is unsettling and dangerous, yet what I have come to understand is that it is too late. Your company’s brand and even an executive’s personal leadership brand are already in the hands of the algorithms that power the AI, and every LLM has its unique algorithms for sourcing information and spitting that back to the requestor. Having differing algorithms across different AI tools makes it difficult to portray a unified corporate brand to the outside world. Each LLM prioritizes different sources of information and uses different language and verbiage to summarize the outcomes. Even the search engines summarize a company’s brand at the top of the search results when a stakeholder seeks information.
It's not to say that search engines prior to AI delivered unified, truthful information back to the requestor. Yet before AI summaries, the “old” way of searching required the requestor to dig through many different links to piece together their own understanding of the company’s brand. As before, we needed to read multiple newspaper articles, books, and magazines to discover the truth about an issue and situation.
The seeking truth function has been outsourced to algorithms we do not write. Now, programmers have developed a process called Truth Discovery that is designed with algorithms to attempt to get to the core truth, yet that discovery process lacks human discernment between what is right and what is wrong.
I therefore define the Truth Algorithm as: human discernment based on judgement, lack of judgement (i.e., eliminating bias), wisdom, human connection, experience, research, reason, inquiry, ethics, and empathy.
For me, this human discernment was summed up in a book I read years ago by Malcom Gladwell, Blink. In this book, Gladwell gets to the core of human intuition. I have not read it in years, but I remember the bottom line about “thin slicing.” When a human makes a gut judgment based on wisdom, intuition, facts, and experience, we should not dismiss it. Most of the time, we are correct and can research to challenge our human bias.
LLMs are not skilled in human intuition and instinct, which C.S. Lewis noted in his book Mere Christianity, is the innate ability of humans to distinguish between right and wrong.
Returning to our Arthur Page discussion on LLMs, I asked the moderator whether LLMs have the “Wiki” or “Snopes” of fact-checking so we could determine whether the AI-generated responses were fact or fact mixed with fiction (or extrapolation). The moderator, who was quite knowledgeable, responded by saying that there is no way to test the truth of brands in LLMs; we can only test for trust in brands.
Woah. No truth in brands in LLMs?
How can a company manage a brand in an information-sharing, gathering, and generating space if there are falsities out there? In fact, during the discussion, I learned that LLMs are only about 22% accurate. Would you want to hang your brand on a 22% accuracy rate? And as consumers, do we want 78% B.S. when we seek to understand the company behind a brand? The answer is “NO” to both of these questions.
How do we get to the TRUTH ALGORITHM, not just the Truth Discovery algorithm for a company’s sustainable brand? Influencing stakeholders' information universe is challenging when managing a company's brand. Yet, it is within a company’s leadership's control to build a brand based on truth and put that truth into the world.
Here are a few recommendations:
No Greenwashing, only “Green Doing.” Sustainable brands are built on sustainable actions, not sustainable marketing. If a company is marketing a sustainable brand and not actually doing the work of changing practices, measuring, monitoring, and reporting, then the sustainable brand is being built on sand, not rock. Truth in action is the first input to the truth algorithm.
Tell the Truth. Truthful communication is a brand builder and a brand saver. In moments of glory and crisis alike, telling the truth about a company’s decisions, actions, impacts builds long term goodwill for the brand, even if in the short term, the truth hurts. Telling the truth demonstrates vulnerability and a willingness to do better.
Update Content. Publish high-quality, optimized content on your website. Search Engine Optimization (SEO) still matters, yet what matters most is consistent, regular, and refreshed content across multiple platforms (e.g., website, social media, etc.). LLMs are pulling from company websites for information on a company, so consider the website ground zero for brand building in LLMs.
Press Releases. Press Releases matter. LLMs seek more “credible” sources of information that have been issued on the newswire, so bring back the press release.
Earned Media. LLMs source from more credible media sources that are not paid advertising. Build genuine relationships with writers and reporters to create opportunities for inclusion in articles, features, executive interviews, and the like.
Wikipedia. Monitor your Wikipedia and ensure it is regularly updated with new developments and scrubbed for inaccurate information. LLMs use Wikipedia as a source of truth.
Utilize Multiple LLMs. Invest in more than one LLM. Algorithms are different, and it is important to get a consolidated understanding of your company’s issues and information swirling out there.
Brand Scan. Conduct a regular scan of your brand on LLMs. Note the sources that are being referenced for summarized results and go down the rabbit holes to explore those sources.
Visibility matters. Keep your executives consistently on message at trade shows, industry roundtables, and prominent conferences where media and other influencers are reporting.
Relentlessly Pursue Sustainability. The more the company is acting and living into sustainability as a way of doing business, the more the storyline shifts in the company’s favor. Your most recent actions are the truth. If the company redirects its focus toward more sustainable practices and the LLM is flooded with the truth of its actions, a more sustainable brand begins to surface.
Let us not forget that brand equity is built on the truth. As Gandhi reminds us, understanding truth is a dynamic process, and the truth may change over time as actions change or as new information is presented.
It is never too late for companies to use truth as the foundation of their sustainable brands. Humans manage brands, and humans buy brands. The truth algorithm is based on human discernment, something an LLM cannot and will never replace.
Brands are built by the humans who live, love, and lead them.
Devry Boughner Vorwerk is Founder and CEO of DevryBV Sustainable Strategies