Observed Signal · Aug 17, 2026 · Opinion · Source: AdExchanger · Impact: 2/5 · Sentiment: Neutral
Deciding What AI Should Never Control in Advertising
Eric Picard of Fluency argues that while LLMs and AI agents will become part of advertising workflows, the industry must clearly define what parts of campaign operations AI must never control. He warns that LLMs are probabilistic and therefore unsuitable for making irreversible budgetary or execution decisions without human-in-the-loop governance. Picard outlines three lessons from running automated advertising at scale: platform APIs do not teach operational wisdom, governance must be encoded into automation from the start, and the most valuable automation scales a buyer’s strategic thinking while keeping deterministic execution for money-moving systems.
Industry-relevant opinion on governance and architecture for AI-driven (agentic) advertising; influences how agencies and buyers design automation but is not a major platform or policy announcement.
Track Fluency Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your Polaris7 workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Opinion piece by Eric Picard (Fluency) published on August 17, 2026.
- The article argues LLMs are probabilistic and should not be allowed to make irreversible advertising spend or execution decisions without human oversight.
- Picard presents three lessons from running automated advertising at scale: API knowledge ≠ operational know-how; automation requires built-in governance; and automation should scale a buyer's strategic intent with deterministic execution.
- The recommended architecture places LLMs at the buyer interface (expressing intent) and enforces deterministic, rule-governed execution layers that cannot be silently circumvented by agents.
Connected Companies & Entities
10 Entities mapped“By Eric Picard, Fluency”
“The answer that seems obvious is to wire up an LLM, give it access to Meta or Google through MCP and put it to work.”
“The answer that seems obvious is to wire up an LLM, give it access to Meta or Google through MCP and put it to work.”
“Access Intelligence, LLC - All Rights Reserved”
“Our revenue growth is below our expectations, The Trade Desk CEO Jeff Green told investors.”
“Pinterest expanded tvScientific CEO and Co-Founder Jason Fairchild’s title to general manager of programmatic and affiliate.”
“Pinterest expanded tvScientific CEO and Co-Founder Jason Fairchild’s title to general manager of programmatic and affiliate.”
“AppLovin's Play To Reach Non-Gaming Advertisers”
“Andrea Kwiatek, director of strategic partnerships at the indie agency Goodway Group, told AdExchanger.”
“Liftoff's Message For Investors During Its First Earnings Call: We're Not Just A Gaming Company”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Digital Advertising Needs AI Guardrails
An AdExchanger opinion piece argues the biggest AI risk in digital advertising is autonomous decision-making without clear ownership, governance or accountability. The author warns that AI can reduce operational friction while increasing systemic, high-impact failures when autonomous systems make pricing, targeting, optimization or creative decisions at machine speed. Citing examples from other industries — Air Canada’s chatbot liability, Zillow’s iBuying losses, and Microsoft’s Tay chatbot — the article outlines a plausible algorithmic pricing-collusion risk and other failure modes (illegal creative, privacy violations, autonomous contracting). It recommends a formal control layer with three core components: configuration authority (clear human ownership), predefined acceptable risk thresholds, and reversibility/auditability (logs, rollback). The piece calls for two operational tracks — bounded experimentation and gated production — and emphasizes retaining experienced human operators to govern AI systems.
Marketers Should Not Let AI Make Decisions
This opinion article argues that marketers should not delegate strategic decisions to AI without the necessary expertise to evaluate the output. The author, with 28 years of experience, advises that AI should support implementation rather than decide positioning, priorities, or investments. The article emphasizes the importance of understanding the business, customer behavior, and sales cycles before accepting AI recommendations. It suggests practical steps such as training staff to review campaigns, asking AI for proof, testing recommendations on a small scale, and maintaining accountability for results. The piece warns against accepting AI suggestions blindly, especially in areas like search marketing, where changes can be far-reaching. The author also discusses the value of AI for ad variations and tagline ideas but not for substantive copywriting. Ultimately, the responsibility for whether marketing works remains with human marketers.
Balancing AI Automation and Human Control in Advertising
Advertisers are increasingly handing creative and optimisation tasks to AI, embedding generative models across the ad lifecycle. An IAB report cited widespread use of generative AI for video creative and buyer intent to adopt it. Big platforms are accelerating automation (Meta announced plans to automate ad processes), while large agency groups are responding: WPP announced a major restructure, asset sales and job cuts aiming for £500m annual savings by 2028. Industry data from the IPA shows a steep workforce decline in the UK ad sector in 2025, with younger staff and creative agencies most affected. The piece argues that human oversight, boundary-setting and intentional use of AI are necessary to preserve brand craft and avoid consumer backlash, citing pulled or criticized AI-generated ads from McDonald’s and Coca-Cola.
Receive all future signals for Fluency & competitors in your briefing
Never miss material M&A deals, leadership changes, account wins, or product rollouts. Polaris7 monitors 20,000+ primary sources 24/7, delivering verified telemetry directly into your executive briefing.