The advertising industry has spent the better part of a decade promising that machines would make everything more efficient. Now, at what appears to be the first annual meeting of an organization devoted to celebrating that promise, attendees found themselves wrestling with the less comfortable corollary: When the machines get it wrong, whose neck is on the block?
The Agentic Advertising Organization, which held its inaugural gathering last week under the name The Foundry, convened executives, academics, and former trade body leaders to discuss so-called agentic systems — artificial intelligence tools that can plan, buy, and measure media campaigns with minimal human involvement. The enthusiasm for such systems, which offer the headcount reductions that have become a recurring feature of agency life, was tempered by a persistent and apparently unwelcome guest at the proceedings: the question of accountability.
"An agent shouldn't evaluate its own work," one workshop spokesperson summarized, articulating what might seem an obvious principle but one that autonomous optimization systems are not inherently designed to honor.
Colleen DeCourcy, chief marketing officer of Sonos and a former creative agency executive, offered a characteristically pointed assessment of where efficiency-obsessed automation tends to lead. "Programmatic is really good at attributing value to people who were already on their way to you," she said, suggesting that the industry had prioritized what could be measured over what might actually persuade anyone of anything.
Of A.I.-generated creative work, Ms. DeCourcy was no gentler. "It can harvest a field," she said. "It cannot plant one."
(One imagines the farming metaphor landed somewhat awkwardly among attendees whose livelihoods depend on the harvesting being automated as rapidly as possible.)
Ruben Schruers, chief executive of the consultancy Ebiquity, raised the legal dimension that few seem eager to resolve. "Is it ChatGPT or Claude who is legally responsible?" he asked. "Or is it the licensing or maintaining the user-agent, or the exchange system on which the agents engage with each other?"
The question, for the moment, remains academic — which is perhaps why it was comfortable enough to discuss at an inaugural meeting. Whether it stays academic is another matter.
Original story published in Digiday: "Ad Tech Briefing: Accountability in adland's agentic era? - Digiday", by Ronan Shields