"Large language models - the ChatGPT class of AI - are welcome almost everywhere except where ad dollars are actually spent. Yes, it sounds contrarian. Planning, buying and optimization autonomously by LLM-powered agents are always framed as just over the horizon. In practice, that switch remains conspicuously off. For now, LLMs are being used as accelerants, not decision makers. They compress workflows. They shrink timelines. They surface insights faster. They do not spend the ad dollars - and that is not an accident."
"At the Consumer Electronics Show in Las Vegas last week, one ad tech trading automation platform cut through the ambient AI enthusiasm with a blunt reality check. It can reduce two-hour workflows to 10 minutes with near-zero errors but stops short of actually spending ad dollars. "At QuantumPath, we want to automate the workflow, not the buying decisions," said its CEO Jeffrey Hirsch."
"Let's start with the most common: the tech. LLMs are designed to operate in open-ended semantic space and sampled probabilistically - a mismatch for programmatic auctions that demand fast, repeatable, deterministic logic. That keeps them parked at the edges of transactions - planning, setup, reporting and analysis - rather than at the crux of it. That boundary, however, is not being treated as permanent."
Large language models are being deployed across advertising to accelerate planning, setup, reporting and analysis while stopping short of executing ad spend. Many platforms and agencies draw a clear boundary between automation that speeds human work and automation that replaces humans at the point of spend. Practical constraints include a mismatch between probabilistic LLM outputs and the deterministic needs of programmatic auctions, imperfect measurement, and unresolved questions of accountability when machines mishandle inventory. Some resistance reflects institutional self-preservation, but ad-tech executives are experimenting and view the current boundary as potentially temporary rather than permanent.
Read at Digiday
Unable to calculate read time
Collection
[
|
...
]