You’re Using AI At The Execution Layer. The Value Is In The Judgment Layer

The instruments are deployed. The licenses are paid. And if you happen to’re a senior SEO or GEO practitioner proper now, you’re most likely utilizing AI daily – for drafts, for summaries, for first passes at content material that used to take twice as lengthy. That’s actual productiveness, and it’s not nothing.

It’s additionally not the return the funding is able to producing. And the hole between what you’re getting and what’s out there isn’t a software drawback. It’s a mode drawback.

A peer-reviewed study published at the 2025 ASIS&T Annual Meeting by Tim Gorichanaz at Drexel College offers that drawback a reputation (h/t to Shari Thurow for pointing me at this paper!). Analyzing 205 real-world ChatGPT use circumstances, Gorichanaz recognized six distinct modes by which individuals truly use AI: Writing, Deciding, Figuring out, Ideating, Speaking, and Critiquing. The info got here from Reddit and skews Anglophone, which limits its generalizability, however the taxonomy it produced maps uncomfortably properly onto how most practitioners are literally working. Two modes dominate. 4 are being left on the desk. The 4 being left are those that decide whether or not AI makes you extra strategically helpful or simply sooner at execution-layer work.

That distinction issues extra proper now than it has at any prior level on this trade’s historical past.

The Two Modes Everybody Defaults To

Writing was the biggest class in Gorichanaz’s knowledge at 47% of noticed use circumstances – drafting, modifying, summarizing, translating, producing. McKinsey’s 2025 State of AI survey confirms this on the enterprise stage: essentially the most generally reported AI use circumstances are content material drafting and data seize, and 63% of organizations utilizing generative AI apply it primarily to create textual content.

Figuring out – explaining one thing, answering a factual query, summarizing a doc – was one other 10% of the research’s knowledge, and represents the opposite pillar most practitioners have constructed their AI workflow round. Analysis a subject, get a synthesis, transfer to the following job.

Collectively, these two modes account for the overwhelming majority of how AI is being utilized by practitioners and enterprises alike. Each have actual worth, but neither is the place the leverage is. And in case your AI apply begins and ends there, you’re utilizing an more and more subtle software to do work that was already being automated – simply sooner and at greater quantity.

The opposite 4 modes (Deciding at 21% of Gorichanaz’s pattern, Ideating at 9%, Speaking at 8%, and Critiquing at 6%) are the place the work turns into irreplaceable. They’re additionally the place virtually no practitioner has constructed a deliberate workflow, as a result of no one handed them one, and the stress to point out fast output has constantly crowded out the area to develop one.

The Selections You’re Nonetheless Making Alone

Within the practitioner’s week, Deciding-mode questions are in all places: which queries even have AI visibility publicity value prioritizing proper now, whether or not a model’s retrieval drawback is a content material structure drawback or a sourcing and sign drawback, learn how to allocate effort throughout a portfolio when each search engine optimisation and GEO want consideration and the budget doesn’t stretch to cover both fully, when to escalate a visibility concern to management versus when to repair it within the work earlier than anybody asks.

Most senior practitioners are at the moment fixing these questions with expertise and instinct. That’s not a failure, as expertise and instinct are genuinely helpful, and no AI replaces them. However AI used intentionally in Deciding mode provides one thing expertise can’t present by itself: a structured pressure-test of the assumptions beneath the choice, utilized earlier than the choice hardens.

That requires greater than a very good query. Deciding mode requires giving the AI the related context (aggressive panorama, present visibility posture, historic efficiency, strategic constraints) after which treating what comes again as a real enter to the choice fairly than a draft to be skimmed and put aside. It requires a workflow that doesn’t but exist in most practitioners’ apply, not as a result of anybody blocked it, however as a result of nobody constructed the time or construction for it both.

The identical McKinsey knowledge makes clear what that hole prices at scale: 88% of organizations use AI, however solely 6% qualify as excessive performers producing significant enterprise-wide impression, and excessive performers are 3.6 instances extra more likely to have fundamentally reworked their workflows fairly than merely deployed instruments into present ones. The sample holds on the practitioner stage. Quicker output from an unreconstructed workflow is just not the identical factor as higher choices from a restructured one.

The Gaps No person Briefed

For search engine optimisation and GEO practitioners, Ideating mode has a selected software that almost all will not be utilizing and most needs to be: mapping the entity and authority gaps the model hasn’t acknowledged but.

What angles of topical authority has the model failed to determine that AI retrieval techniques are at the moment filling from different sources? What group indicators (discussion board discussions, aggregated critiques, third-party commentary) are shaping how LLMs represent the brand in response to class queries, and what wouldn’t it take to shift them? What framings of the model exist in mannequin coaching knowledge that the model’s personal content material has by no means addressed or countered?

These are genuinely Ideating-mode questions. They’re additionally questions most practitioners have some model of at the back of their thoughts with no structured methodology for surfacing the solutions. AI utilized in Ideating mode, not “give me 5 content material concepts” however a real iterative exploration with deliberate constraints and actual willingness to observe the output someplace the staff hasn’t already been, is likely one of the most direct strategies out there for locating these gaps earlier than a competitor or a consumer audit finds them first.

The barrier isn’t functionality. It’s the distinction between a Writing immediate with an inventory output and an precise Ideating session. The primary takes two minutes. The second takes twenty, requires a unique posture towards the software, and produces one thing that may’t be replicated by anybody who didn’t do it. That asymmetry is the place practitioner worth will get constructed within the present setting, and most practitioners will not be claiming it.

The Sincere Learn Your Workforce Gained’t Give You

That is the mode with essentially the most direct software to day by day apply and essentially the most organizational resistance, as a result of it requires utilizing AI to search out issues in work the practitioner or their staff has already invested in.

Used correctly, Critiquing is how a senior practitioner catches what inner assessment missed. The weak entity declare in a content material technique that sounds authoritative however isn’t backed by the sourcing AI retrieval systems actually trust. The hole between what the model says about itself throughout owned properties and what a well-prompted LLM surfaces when requested a class query the model ought to personal. The assumed premise in a GEO advice that made sense six months in the past and is now contradicted by how retrieval patterns have shifted.

That final software is just not summary. Operating your individual model (or a consumer’s model) by way of a structured AI Critiquing session earlier than the following technique cycle is strictly the form of proactive work that separates practitioners working on the judgment layer from practitioners working on the manufacturing layer. It’s additionally the form of work that modifications the dialog with a consumer or a management staff, since you’re surfacing issues earlier than they develop into seen within the knowledge fairly than explaining them after the actual fact.

The rationale Critiquing is underused isn’t a governance drawback. It’s a disposition drawback. Organizations and practitioners have broadly educated themselves to make use of AI to provide output, to not interrogate it. Reversing that behavior is a alternative, and it’s one of many extra consequential decisions out there to a senior practitioner proper now.

Rehearsal

The Speaking mode in Gorichanaz’s taxonomy covers AI as a dialog companion, and for practitioners, essentially the most helpful model of that’s rehearsal for the inner and consumer conversations the place the stakes are actual.

The consumer name the place it’s important to clarify why organic traffic is down 30% while AI search visibility is also poor, and it is advisable to maintain two separate causal explanations concurrently with out letting them collapse right into a single narrative that oversimplifies each. The inner briefing the place it’s important to make the case for GEO funding alongside present search engine optimisation finances to a management staff that also conflates the 2 disciplines and desires a single quantity that explains the ROI of each. The company or vendor assessment the place it is advisable to push again on a really helpful strategy with out dropping the connection.

These conversations are recurring and high-stakes, and most practitioners stroll into them with solely their very own psychological rehearsal as preparation. Speaking mode (role-playing the pushback, asking the AI to argue the opposite facet, working by way of the model of the dialog that goes flawed) is just not a alternative for expertise. It’s a preparation methodology that prices twenty minutes and materially modifications the standard of the practitioner who walks into the room.

It doesn’t produce an artifact. It doesn’t present up in a utilization report. EY’s 2025 Work Reimagined Survey, which lined 15,000 workers and 1,500 employers throughout 29 nations, discovered that 88% of workers use AI at work, however solely 5% use it in ways in which basically rework what they produce. The rationale that hole is so extensive is nearly actually that the superior modes – Critiquing, Deciding, Speaking – don’t produce one thing measurable within the second. They produce a greater practitioner over time, which is a return that compounds and doesn’t seem in a dashboard.

What Mode You’re In Is What Layer You’re On

The six-mode taxonomy maps virtually precisely onto the cut up between execution-layer work and judgment-layer work. Writing and Figuring out are execution-layer modes. They’re helpful, they’re seen, they usually’re more and more the modes that AI handles with much less and fewer human involvement. Deciding, Ideating, Critiquing, and Speaking are judgment-layer modes. They’re the place the practitioner’s irreplaceability lives.

A senior search engine optimisation or GEO practitioner who makes use of AI solely in Writing and Figuring out mode is, functionally, positioning themselves as an execution-layer employee at precisely the second when AI is most aggressively compressing that layer. That’s not a prediction about job displacement. It’s an statement about skilled differentiation. The practitioners constructing sturdy worth on this setting are those utilizing AI to make their judgment higher, not simply their output sooner.

Gorichanaz’s research reframes what data want truly means within the AI period, not simply question-answering or uncertainty discount, however what the authors name skillfully coping on this planet, which means the continuing software of sensible intelligence to conditions requiring each understanding and motion. For a senior practitioner, that framing is a helpful diagnostic. The query isn’t what AI can do. It’s which elements of your work require the form of sensible intelligence that compounds with expertise, and whether or not your present AI apply is making that intelligence sharper or simply making every little thing round it transfer sooner.

McKinsey’s workplace research finds that just one% of leaders name their corporations mature on AI deployment, which means AI is absolutely built-in into workflows and driving substantial enterprise outcomes. The practitioner-level model of that hole is simply as extensive, and simply as fixable.

If you happen to mapped your precise AI utilization in opposition to the six modes this week (not what you plan to do, what you truly did), how would the distribution look? How a lot was Writing and Figuring out? How a lot was Deciding, Ideating, Critiquing, Speaking?

The practitioners who shut that hole intentionally, who construct even a minimal workflow across the judgment-layer modes, will not be doing one thing unique. They’re doing one thing most of their friends will not be. In a self-discipline the place the execution layer is getting compressed by the identical instruments everybody has entry to, that hole is the one value closing first.

To see what I simply constructed after months of labor, you’ll be able to learn extra about data for decisions and evidence for your conversations.

Extra Sources

This put up was initially revealed on Duane Forrester Decodes.

Featured Picture: Roman Samborskyi/Shutterstock