Advertisers hungry for AI may be to blame for its increased risks

When Grant Gudgel signed a recent open letter calling for a temporary pause to advanced AI development in order to better assess the risks, he wasn’t actually supporting a hiatus. The founder of Jetson Advertising Intelligence, an ad tech platform, Gudgel regularly uses AI and has integrated the technology as a primary component of his business. Yet he still has concerns about the breakneck speed at which AI is growing, and wanted to call attention to these issues. In particular, he is worried about AI-generated content, which he views will only become more difficult to discern from human-generated content over the next three years. Stephan Pretorius, WPP’s chief technology officer, also signed the letter, yet he remains confident that the enterprise use cases of AI are only going to increase and improve, he told Ad Age. These ambivalent positions exemplify a conflict stirring in the ad industry, in which marketers are worried over too eagerly adopting nascent AI technology. At the same time, however, this adoption is happening in full force, led by agencies racing to launch new tools built with generative AI capabilities. Omnicom, WPP, MNTN and Accenture are just a few of the players to have announced such services in recent months. But while new products roll out each week, the risk potential for issues like copyright infringement, data privacy compliance and the spreading of misinformation is not going away. In fact, these very advertisers may be fueling it. Opinion: How Gen Z is embracing AI in the workplace “As a society, we’re not willing to restrain industry from providing services,” said Lisa Anthony, an associate professor in the department of computer and information science and engineering at the University of Florida with a Ph.D. in Human-Computer Interaction. A tension thus emerges, she said, between being the first to market and recognizing that a profit-driven motive is not the most responsible force to push AI development forward. Advertisers are caught in the middle of this tension, compelled to provide clients the most up-to-date tools, especially as competitors seek to do the same. Or, as Gudgel put it: “If I don’t, somebody else will.” AI marketing blog How brands are using ChatGPT, DALL-E 2 and other tools Read more here The past eight months suggest that more AI development, as opposed to less, may only perpetuate this arms race, eventually to a point where there may no longer be ample time nor space to create preventative measures. Recall that marketers had already been using AI technology for over a decade, yet with the onset of generative AI through bots such as ChatGPT, almost any notion of forbearance went out the window. As the capabilities of generative AI become stronger and more seductive, there is little reason to believe that marketers will not pounce on those offerings, too. “We could wish that it would slow down, but it’s not going to,” Gudgel said. Recent news: ChatGPT’s brand success is hardly artificial Perpetuating risks Marketers who spoke with Ad Age generally agreed that the risks posed by AI can be grouped into two categories: Legal risks are those that could have immediate consequences on companies due to existing policies, such as copyright infringement, data privacy noncompliance and the dissemination of inappropriate content. Social risks are broader and more severe; these include the spreading of misinformation, excessive job replacement by AI and even existential-level threats to humanity, as highlighted in a recent open letter by the Center for AI Safety, a non-profit organization. The recent avalanche of generative AI-powered advertising tools mostly raises flags in the legal category. For example, the copyright implications of AI-generated content are still very much up in the air. Some media companies, such as Adobe and Shutterstock, have built precautionary measures into their AI tools, but a wider consensus on proper conduct—including actual legislation—has not yet been established. Issues such as copyright and data privacy, which contributed to a temporary ban of ChatGPT in Italy earlier this year, are made murkier by the fact that even if an AI-powered tool is behaving properly at the product level, it may be behaving improperly at the foundation level. For example, AI developers including OpenAI and Stability AI have already been hit with numerous lawsuits alleging their popular models are copyright or privacy non-compliant. This means that an agency using their models in its ad tool could be fueling improper actions, regardless of whether or not it introduced problems via its own data. Related news: AI copyright ruling could pose issues for agencies “Hasty implementation is risky,” said Uri Lichter, CEO of ad tech company Intango. Given that the legal implications of generative AI-powered technology are still not well understood, Lichter views over-promising value as another risk marketers need to consider. In press releases, agencies often embellish the disruptive capacity of their AI tools. They opt for the kind of phrasing meant to drive hype, which is typically harmless, unless real risks are involved. Of course, generative AI really does offer notable benefits, but this is not a license for throwing caution to the wind. One possible strategy that Lichter posed would be keeping clients in the loop about your AI development, but at the same time slowing down your wide-scale, market-ready implementation of the technology. In addition to the intricate legal risks of deploying AI, there is growing evidence that social risks are becoming a bigger threat. Some 30,000 ad industry jobs are projected to be wiped out by AI in the next seven years, according to a recent report from Forrester Research. While some may interpret the displacement as relatively small, Forrester’s number depends on major questions currently pertaining to AI—such as regulation, copyright and accuracy—being answered within the next two years. Should these questions remain unanswered, perhaps due to AI development moving faster than policymakers, all bets could be off. Bias in media planning is another social risk that advertisers have long struggled with, and it could be exacerbated through the automaticity of AI, said John Roescher, founder and CEO of design firm Raw Materials. “Good strategists and creatives working in shops with systems-thinking rigor and empathy-based practices can correct these trends,” he said, “but using AI to analyze, ideate, create and optimize stands to perpetuate them.” Agencies push back Soon after WPP’s Pretorius signed the open letter calling for a pause to AI development, WPP announced a new generative AI ad platform in partnership with Nvidia. Pretorius does not see this as hypocritical. “Signing that letter does not mean I am anti-AI or anti-generative AI,” he told Ad Age. Focusing on the application of existing AI technology, as opposed to deploying new forms of AI, fits within the remit of the open letter’s mission, he said. Other agency leaders who spoke with Ad Age also pushed back on the concern that their fast-acting strategies are perpetuating the risks posed by AI. “We shouldn’t slow down, but rather speed up in figuring this out,” said Paolo Yuvienco, chief technology officer at Omnicom. Within the last 30 days, the marketing giant has struck deals with Google, Microsoft and Amazon Web Services to bring generative AI to its Omni data platform. The lack of a regulatory framework is not surprising, said Yuvienco, and should in fact compel advertisers to push forward with development so that they can more quickly learn better and safer practices. “We’re always ahead of regulation,” he said. “It’s our responsibility to stay ahead of this.” Jatinder Singh, global head of data and analytics for Accenture Song, echoed this point. Accenture Song has a designated team for hashing out responsible AI use, which “enables us to move forward and mitigate the risks at the same time,” he said. Accenture (its parent company) announced last month a $3 billion investment in AI opportunities, including a platform that uses generative AI to determine how AI can fit into clients’ businesses. More news: Inside agencies’ AI practices Some agency leaders, however, are concerned by the Wild West-like vibe of the AI space, and are taking an actual pause to assess risk versus gain, said Grant Theron, CEO of Commonwealth//McCann and global business lead of General Motors at IPG. One particular issue raised by Theron is the possibility that a brand uses AI to unintentionally infringe upon copyright, which could then transfer risk back to the agency due to the agency-client agreement. A better approach is to use AI in closed environments where agencies have control of the data and source materials, he said. Roescher of Raw Materials, who flagged the issue of AI bots worsening bias in media planning, paints the predicament in simple words. “There's so much we don't know about how they work and why they create what they create,” he said. “It's hard to truly defend decisions and creative output if some part of it is an enigma.” Sign up for Ad Age newsletters From influencer marketing to agencies, get the latest news and analysis delivered to your inbox Click here

Advertisers hungry for AI may be to blame for its increased risks

Featured Stories