The biggest AI marketing mistakes and how brands can avoid them

Google sparked outrage last month when its Gemini conversational bot went off the rails—just the latest instance of AI acting out and leaving egg on the face of a notable brand. But these incidents, which have only become more widespread as technologists and marketers push generative AI to the public, are also learning opportunities for brands hoping to avoid similar outcomes. The importance of establishing robust guardrails and reconsidering how AI mixes with social issues can be gleaned from the mistakes made over the past year. Marketing experts say this is just the beginning. Be it Google or Levi’s, Meta or Chevrolet, companies diving into AI are exposing themselves to risk that is increasing with every new AI tool that hits the market. “The story of the next six months is that we’re going to see the limits of AI more than last year,” said Adam Simon, managing director of IPG Media Lab. Rogue chatbots Chatbots that go off the rails are the most common—and highest-profile—mistakes seen in the AI space. Tech giants have typically borne the brunt of this issue given their tendency to tout new and imperfect platforms to the public. Last month, Google’s Gemini conversational bot overcorrected for sensitivity by generating ahistorical images and refusing innocuous queries. The malfunction caused an uproar on social media, and advertisers lamented the potential erosion of trust that could follow the incident. Microsoft also landed in hot water when its Bing chatbot went haywire in countless interactions early last year. From urging users to leave their partners to criticizing them for being “unreasonable”, the system proved how risky it is for providers to release their bots to the public. Meta received backlash when some of the celebrity personalities tied to its chatbots denied their real-world brand partnerships, and a few denounced the brands themselves, according to Fortune. Rogue bots have caused headaches for several brands, too. A customer service assistant for U.K.-based delivery service Dynamic Parcel Distribution (DPD) swore at a user and produced a critical poem about DPD; a bot serving Chevrolet agreed to sell a user a vehicle for $1—with “no takesies backsies;” and an AI platform offered by the National Eating Disorders Association gave inappropriate weight loss advice, per The Wall Street Journal. Parcel delivery firm DPD have replaced their customer service chat with an AI robot thing. It’s utterly useless at answering any queries, and when asked, it happily produced a poem about how terrible they are as a company. It also swore at me. 😂 pic.twitter.com/vjWlrIP3wn— Ashley Beauchamp (@ashbeauchamp) January 18, 2024 Marketers deploying AI chatbots should err on the side of caution when establishing guardrails, said Andrew Frank, research VP and distinguished analyst at Gartner. This means staying focused. So whatever the primary purpose is of an AI assistant—from generating fun recipes to providing more information on a specific product—should probably be the only capability for that service, as a way to prevent it from venturing into uncharted territory. Also read: Burger King CMO on AI Avocados From Mexico’s guac recipe generator, for example, rejected certain inputs for recipe ideas that could result in unsafe moments for the brand. Even if a user wanted to generate, say, a cigarette-flavored guac, they were unable to do so through the brand’s platform. Volvo’s bot on WhatsApp was programmed to answer only questions about a certain vehicle; if a user asked an irrelevant question, the bot would decline a response or offer to hand off the user to a human customer care specialist. “A lot of brands are finding the open-ended chatbot [to be] challenging and not ready for public-facing activations,” Frank said. Check out the latest Ad Age events and award programs Social issues and transparency Brands are attracted to purpose-driven marketing, but history shows that AI may confuse these efforts more than convenience them. Levi’s drew ire last year for announcing it would AI-generate diverse body types for modeling purposes. The intent may have come from the right place, but that didn’t stop hordes of people from calling out the brand for cutting corners around inclusivity. So instead of increasing diversity by hiring actual Black and Brown models, Levi's is going to get AI models. The conversation around diversity has hit the "jumped the shark" level. Cuz WTF. https://t.co/kdZvMzTtOL— Luvvie (@Luvvie) March 27, 2023 During this year’s Super Bowl, a spot from the “He Gets Us” campaign confused many viewers who thought the creative was AI-generated, when in fact it was produced by a human artist. To make matters worse, some viewers lashed out for an unwanted tie-up between AI and Jesus. Both missteps evince the sensitivity that many consumers have around serious issues. “It doesn’t necessarily mean that serious themes should be off-limits for AI, but rather that brands need to be especially mindful of the context, presentation and the potential for misinterpretation,” Bernard Marr, a technologist, founder and chief executive of consulting firm Bernard Marr and Co, previously told Ad Age. The need for transparency is also a message that emerges from these mistakes. In another recent instance, Instacart was using AI to generate images for its website. Reddit users users began compiling some of the strangest pictures, including two chickens conjoined together and a hot dog with the interior of a tomato. The platform removed many of the images sometime after, without a statement as to why and what fueled the use of AI in the first place. To prevent these mess-ups, a human-driven checks and balances system is critical, said Rio Longacre, managing director of advertising and marketing transformation for Slalom Consulting. “People looking for shortcuts is where we start to see issues,” he said.

The biggest AI marketing mistakes and how brands can avoid them

Featured Stories