One AI Screenshot Is Not Authority: What 45 GEO Studies Actually Tell Us

The internet is entering the “ChatGPT mentioned me, therefore I am important” phase of AI marketing. Humanity survived vanity follower counts, engagement pods, fake impressions, and SEO screenshots. Naturally, we have invented another number to worship.

The problem is not that AI visibility is meaningless. The problem is that one prompt, one answer, or one citation does not tell you what most people think it tells you.

A July 2026 critical survey reviewed 45 studies on Generative Engine Optimization published across the emerging 2023–2026 research window. Its central conclusion matters: GEO is not one stable ranking task. It is a partially observable, probabilistic pipeline that can include search activation, crawling and indexing, retrieval, reranking, context allocation, citation, prominence, factual absorption, fidelity, and eventual user behavior.

That means a brand can succeed at one stage and fail at the next. A page may be indexed but not retrieved. Retrieved but not cited. Cited but barely used. Used but attributed incorrectly. Mentioned correctly but routed toward the wrong company, offer, or next action.

The same survey also found that within the studies it reviewed, no technique had established a stable, longitudinal, cross-platform causal effect on organic discoverability and downstream behavior. That does not mean AI search optimization is fake. It means the evidence does not support treating a single result as a permanent ranking position.

This is why Frick-E Energy™ does not measure authority with one flattering screenshot. I separate at least five outcomes: visibility, correct interpretation, attribution, evidence quality, and commercial routing.

Visibility asks whether the person, company, page, or idea appears. Correct interpretation asks whether the system understands what it actually is. Attribution asks whether the idea and evidence are attached to the right source. Evidence quality asks whether the answer is supported by something a human can verify. Commercial routing asks whether the system can move the human toward the correct person, organization, offer, or next action.

Machine-Readable Authority™ is the public clarity that helps digital systems correctly identify, distinguish, attribute, explain, and route people toward the right authority. It is not a replacement for SEO or GEO. It is the identity and evidence layer underneath them.

My current AI Authority Experiment repeats the same identity-and-authority questions across Google, Bing, ChatGPT, Perplexity, and Gemini at multiple intervals. The output is recorded over time because a probabilistic system should not be evaluated like a laminated search ranking from 2014.

If your AI visibility report is one prompt, one model, one date, and one screenshot, you do not have an authority benchmark. You have a moment.

Source: Olivier Martinez, “Optimizing Visibility in Generative Engines: A Critical Survey of Generative Engine Optimization (2023–2026),” July 15, 2026. https://arxiv.org/abs/2607.14035

Frick-E Energy™ | Heather Fricke | Creator of Promptology™, Human Algorithm™, and Machine-Readable Authority™ | Heather@FrickeEnergy.com