AD FIELD NOTESAI advertising · the work behind the output

Research library / September 2026 notebook

The field in one sitting

A concise map of what AI changes in advertising—and what still needs human judgment.

Artificial intelligence is changing advertising in three different ways at once:

  1. It is becoming a new creative material for images, video, sound, voices, characters, and interactive experiences.
  2. It is becoming a workflow layer that helps teams research, brief, storyboard, version, translate, test, and publish.
  3. It is becoming a media and customer layer that decides who sees which message, where, when, and sometimes what happens next.

The most important distinction is not between “AI ads” and “human ads.” It is between work where AI expands a clearly human point of view and work where the tool quietly replaces the point of view with generic output. A prompt can produce an image; it cannot establish why the image should exist, whether the product is represented truthfully, whether the likeness is authorized, or whether the result is worth putting in front of people.

The evidence supports a practical, conditional view:

  • Speed and volume are real benefits, especially for exploration, resizing, rough visualization, copy variations, translation drafts, editing, and routine campaign operations. Google, Meta, TikTok, Amazon, Adobe, Spotify, and other platforms are embedding these capabilities directly into ad workflows.12345
  • Quality and effectiveness are task-dependent. A Harvard/BCG field experiment found strong gains on tasks inside the model’s capability frontier but worse performance on a strategy task outside it. A customer-support field study found a 14% productivity lift, concentrated among less experienced workers. A later METR randomized study found experienced open-source developers took 19% longer with early-2025 AI tools in its setting.678
  • Most campaign performance claims are vendor or agency claims, not independent causal evidence. Reported lifts can be useful leads, but they should not be presented as proof that AI-generated creative is generally more effective than human-created creative.
  • The real cost moves upstream and downstream. Prompting and generation may be cheap; strategy, art direction, rights, consent, legal review, factual verification, cleanup, localization, accessibility, production management, and measurement remain paid work.
  • Disclosure and provenance are becoming production requirements. The EU AI Act’s transparency rules apply from August 2, 2026, including labeling certain AI-generated or altered content and telling people when they interact with AI. The FTC continues to apply ordinary truth-in-advertising and endorsement principles, including to AI-generated testimonials and impersonation. C2PA Content Credentials provide a technical way to record provenance, but they do not replace legal clearance or audience-facing disclosure.91011

The recommended operating model is human-directed, AI-assisted, reviewable, and traceable. Use AI to widen the option set and reduce repetitive production work. Keep the brief, taste, cultural judgment, consent, claims, final edit, and accountability with named people.

Follow the evidence

This chapter comes from the September research notebook. Linked sources and qualifications remain attached to the claims.

Open all sources and footnotes →

The weekly field note

One useful note for the next brief.

Campaigns worth studying, the work behind them, and questions to bring into your next creative review. Delivered weekly.

How we handle your email