Talk:Artificial intelligence optimization

I think this article is TOOSOON and that the SEO article could absorb it. Dflovett (talk) 16:26, 28 July 2025 (UTC)Reply

Following up on my last note: I've attempted to make some edits to this article, including deleting some sections. It still needs a lot of work and I'm not certain it should exist outside the SEO article. Dflovett (talk) 07:34, 20 August 2025 (UTC)Reply

Both GEO and AIVO are unrelated to SEO. Instead, they are situated within the broader field of 'AI Search', a developing discipline focused on how large language models retrieve, rank, and generate information. Derosse (talk) 17:57, 21 August 2025 (UTC)Reply

I'm very familiar with both SEO and the developing field of GEO/AIVO/Whatever you want to call it. They are related to SEO. I also think you agree with me, because you attempted to add a few paragraphs about GEO/AIVO etc to the SEO article. SEO, GEO, AIVO all fall under the same umbrella. We just don't know what to call the things under that umbrella yet. Dflovett (talk) 02:33, 23 August 2025 (UTC)Reply

David as you are affiliated with Rocket 55, an SEO agency you are not an impartial judge in these matters. SEO is not the umbrella term for AI search, although I fully understand your desire to make it so. In the interests of fairness and impartiality these matters should be adjudicated by individuals with no "skin in the game". Derosse (talk) 05:03, 23 August 2025 (UTC)Reply

I'm happy to sit out of future edits on these articles but these articles do need to be edited by people with familiarity with the matter. Here's the response I shared with you on the LinkedIn post you wrote about me:

I read your Medium post and this post. Ironically, I think you and I agree a lot more than you realize. SEO and AI Search are not the same thing. Related, yes, but not the same. In my opinion, SEO should not be the umbrella term for AI Search Optimization or whatever it ends up being called. I have tried to get the articles for GEO and AIO on Wikipedia cleaner. I also think the SEO article has always been a mess and remains a mess. AIVO is one of the better terms for it I've seen. The challenge we're all facing is that Wikipedia is not the shaper of industry terms but the reflection of them. Third-party coverage is what will ultimately determine what Wikipedia calls it. Dflovett (talk) 07:02, 23 August 2025 (UTC)Reply

It is surprising that you would call someone else out for a conflict of interest when you have been writing about your own work and linking to your own sites in violation of Wikipedia's guidelines, Derosse. MrOllie (talk) 13:54, 23 August 2025 (UTC)Reply

It's surprising that you would call me out on this, when I've already filed a COI.I will be charitable and assumed you've just missed it, rather than deliberately spreading FUD. Derosse (talk) 14:00, 23 August 2025 (UTC)Reply

In any event, AIVOStandard.org is a non profit entity so it is not "my work". It's a public, open standard. Not a partisan effort like much of the discourse in SEO. Derosse (talk) 14:18, 23 August 2025 (UTC)Reply

That you 'filed a COI' does not mean you can ignore every other policy Wikipedia has. Edits like this one are considered to be linkspam by the Wikipedia community. MrOllie (talk) 14:19, 23 August 2025 (UTC)Reply

Have you disclosed a COI in this matter, MrOllie? Do you work in the SEO sector or have connections with it? Derosse (talk) 14:21, 23 August 2025 (UTC)Reply

No, I don't. Just a long time Wikipedia editor who regularly deals with issues of linkspam and COI editing. MrOllie (talk) 14:22, 23 August 2025 (UTC)Reply

I note that you are the subject of multiple public criticism's and that by 2021 you had made over 85,000 edits, allegedly using algorithms and bots.This would tend to incriminate you as a serial complainer who derives sadistic-like pleasure from interloping in multiple subjects and Talks. Derosse (talk) 14:28, 23 August 2025 (UTC)Reply

Yes, linkspammers don't like me very much and they tend to complain - because I don't allow them to linkspam. You should look at what you're doing and consider if you really want to join that group. Also, have a look at WP:NPA - what you're writing here (as well as the off-site harrassment of Dflovett) are blockable behaviours themselves on Wikipedia. Kindly make some effort to respect the norms of the community you've come into here. MrOllie (talk) 14:33, 23 August 2025 (UTC)Reply

Let's place this conversation in the public domain where you can't hide behind Wikipedia's "Wizard of Oz" curtains. I haven't harassed anyone. Ask Mr.Lovett. It's high time that Editors like you were held to account and named and shamed. Derosse (talk) 14:40, 23 August 2025 (UTC)Reply

Ok, I'll bring it to the administrator's noticeboard and we'll see what they have to say about it. MrOllie (talk) 14:46, 23 August 2025 (UTC)Reply

Go ahead my friend.I wish you well. Derosse (talk) 14:47, 23 August 2025 (UTC)Reply

I, for one, think that the world of AIO whilst (obviously) being related to SEO will diverge rapidly as a discipline, although SEOs are well placed. the whole conversation around tokenization and also the query fan out are two elements of a rapidly evolving ecosystem that lives increasingly independently. If the two are merged, then the SEO page, which is already long and ALSO apparently contains multiple issues, I just think you will have one big article that nobody can agree on, rather than two which can be addressed individually. Now - merging the orphaned content at Generative engine optimization into Artificial Intelligence Optmization seems sensible. That age does not have a "multiple issues" warning (although there probably should be...) but it is short and seem to get confused with Local SEO.

Anyway... I don't want to touch anything as I always get told off when I do. Just adding my thoughts as requested. Dhjv (talk) 16:01, 13 October 2025 (UTC)Reply

I agree to merge GEO and AIO. Dflovett (talk) 12:08, 11 December 2025 (UTC)Reply

Generative Engine Optimization (GEO)is legacy as a result of the archiving of the GEO Standard at GeoStandard.org. It has been superced by Artificial Intelligence Visibility Optimization (AIVO). Derosse (talk) 17:52, 21 August 2025 (UTC)Reply

I agree that there are aspects of this article that are incorrect and need updating. But the AIVO Standard needs more third party coverage before you can credibly make the claim that it has superseded the concept of GEO or the other ideas in this (flawed) article. Dflovett (talk) 02:36, 23 August 2025 (UTC)Reply

The content of this article cites the AIO is also known as answer engine optimization or AEO. That doesn't seem factual. AIO as the article presents is more the discipline of optimizing AI output of LLMs. AEO in the public domain is the newer discipline that brands/companies use to optimize their visibility in AI answer engines, much like SEO (as has been discussed in this talk). It seems these two distinct concepts are being conflated together in this article and that they likely should be separate articles. Unless I'm missing something. Team aivo (talk) 20:36, 24 October 2025 (UTC)Reply

In general, I think these concepts are conflated, which is why it makes sense to mention both of them. I'm not clear on what AIO is and I don't think anyone else is clear on it either. Is AIO practiced by creators of AI systems or people trying to game the AI systems? We currently lack consensus on this. Dflovett (talk) 14:48, 11 February 2026 (UTC)Reply

Hi all — disclosure up front: I’m a developer affiliated with Fabled Sky Research (FSR), one of the groups involved in work sometimes described as an “AIO standards framework.” Because of that conflict of interest, I don’t intend to edit the article or propose specific wording for the mainspace. I’m posting here only to provide technical context, in case it’s useful to independent editors who are evaluating how—or whether—this topic should be represented at this time.

On terminology: “Artificial Intelligence Optimization (AIO),” as used in the materials we published, is not intended as a synonym for “Answer Engine Optimization (AEO).” AEO is generally used to describe brand or marketing practices aimed at improving visibility in AI answer engines, often discussed alongside SEO. By contrast, the AIO framework was conceived as a more evaluation- or measurement-oriented approach, focused on how content and its structure perform within large language model–based retrieval, ranking, and generation pipelines, rather than as a commercial service model.

From a systems perspective, one property of frameworks like AIO relates to how they interact with computational efficiency in LLM pipelines. Modern generative AI systems routinely handle recurring or semantically overlapping queries and are therefore designed to reuse representations through mechanisms such as embedding reuse, retrieval caches, ranked reference stores, and other inference-time optimizations.

Content that is factually stable, structurally consistent, and semantically unambiguous is more amenable to these mechanisms. When information can be reliably retrieved, referenced, and validated without repeated reprocessing, systems tend to incur lower token usage and less redundant computation. This effect is independent of marketing or visibility goals and follows from how LLM-based retrieval, ranking, and generation systems are architected at scale.

As a result, standards or frameworks that emphasize clarity, provenance, and internal consistency can have secondary effects on efficiency—particularly in enterprise or higher-reasoning contexts where compute and inference cost are dominant constraints. This isn’t meant as a claim of environmental benefit, but rather as an architectural consequence of aligning content with how modern AI systems manage reuse, caching, and evaluation.

On commercial intent: I noticed comments suggesting that Fabled Sky or FSR “sells these services.” I’m not here to speculate about intent, but this appears to stem from conflation between different efforts. The AIO materials themselves were published as a free, open, standards-oriented framework and were not positioned as an SEO-style client service.

Separately, there is a distinct project—Objectivity AI—that applies AIO-aligned principles in a commercial or enterprise implementation context (for example, packaged models, dedicated infrastructure, or API-based systems). That project should be understood as an implementation layer that adopts or builds on certain ideas, rather than as the AIO framework itself. Some of the confusion around commercial intent may arise from this distinction not being explicit, perhaps?

At a conceptual level, one of the design goals of the AIO framework was to avoid traditional notions of “gaming” that emerged in SEO. Contemporary LLM-based systems do not rely on narrow ranking signals or static rule sets; instead, they typically involve multi-stage pipelines that may include retrieval, query expansion or fan-out, contextual ranking, synthesis, and post-generation evaluation. In such systems, narrow optimization tactics tend to be brittle and often degrade performance over time.

What tends to improve outcomes consistently is alignment rather than manipulation: producing content that is factually reliable, well-sourced, internally consistent, clearly structured, and semantically unambiguous, with sufficient context for models to interpret provenance and intent. In that sense, attempting to “game” such a framework largely collapses into doing the underlying work well, rather than exploiting shortcuts. While there may be superficial similarities to SEO or generative engine optimization in terms of visibility, the underlying mechanics are meaningfully different.

On sourcing: I noticed that aio.fabledsky.com was removed in an earlier edit, with the explanation that Fabled Sky Research appeared to sell AIO-related services. I didn’t add that source myself, but wanted to clarify the context. The AIO materials published by Fabled Sky were intended as primary documentation of the framework’s definitions and scope, not as a commercial service offering. Any perceived conflict appears to stem from conflation with separate projects that use similar ideas in a commercial context, rather than from the AIO framework itself.

Whether or not aio.fabledsky.com is appropriate to include—and in what capacity—is, of course, for independent editors to decide. I’m only noting this distinction in case it’s useful for anyone reassessing the sourcing or structure of the article.

More generally, I agree with the sentiment expressed elsewhere on this Talk page that the broader field is evolving quickly and that terminology may shift as independent coverage matures. There’s also a risk of the space becoming muddled as commercial incentives increase, similar to what happened historically with SEO. Whatever structure or naming Wikipedia ultimately adopts, it may be helpful to clearly distinguish marketing practices from standards- or measurement-oriented frameworks, and to avoid treating AIO, AEO, SEO, GEO, and AIVO as interchangeable terms unless reliable sources explicitly do so.

I’m happy to answer factual or technical questions about what the framework does or does not claim, but I’ll leave any decisions about article content to independent editors. FSR-Rene (talk) 22:35, 16 January 2026 (UTC)Reply

Did you use an LLM to write this? Dflovett (talk) 14:47, 11 February 2026 (UTC)Reply

Yes. Voice to text and LLM passthrough for clarity and structure. Not sure why it would be an issue. FSR-Rene (talk) 19:58, 13 February 2026 (UTC)Reply

FYI please don't use my comment above as a "primary source" for this material. I was just talking out loud if that's what you meant. There is a lot of talk on this topic but there is no agreed upon consensus of what it is called in the "AI community" (I.E. We call it the "AIO Standards Framework", but this is not some specific global AI concept, at least as of yet. There is too much uncertainty right now in how the big tech players are ultimately going to incorporate advertising into their LLM user pipelines, so this field will most likely be focused on internal systems, documentation, and situations where accuracy and reliability are key. Not for finding the best donut shop in your city, if that makes sense. FSR-Rene (talk) 20:16, 13 February 2026 (UTC)Reply

I'm following you. I think a lot of this is TOOSOON. I think your standards seem fair but they aren't universally accepted standards or "industry standards," at least not yet. Dflovett (talk) 15:05, 16 February 2026 (UTC)Reply

This article was way too long before. Now it's very short, but all the references seem to have survived. I'm considering taking on the task of dropping some of these references. Anyone else is welcome to join me. Dflovett (talk) 14:49, 11 February 2026 (UTC)Reply

I think this is the basic question that needs to be answered by this article, if it will continue to exist. If this is about third parties "optimizing" AI platforms, i.e. trying to manipulate results, then this is not different from GEO or AEO, regardless of what approaches and systems are used. If this is about engineers attempting to improve AI systems, then this is a distinct concept. Dflovett (talk) 12:09, 13 February 2026 (UTC)Reply

Right now I am seeing a lot more of the former. The goal for the framework is more in line with your latter statement (engineers attempting to improve AI systems). We never wrote the standards with the intent they would become a Wikipedia article. Personally, I think it is too soon as this field is still evolving, and the SEO/GEO/AEO/ETC crowd wastes no time referencing our standards as some sort of marketing ploy. We do not offer this as a commercial service and do not endorse any company claiming to use these standards as a marketing tool. That is how this article was first brought to my attention. With that said, all I ask is if this article does stay, the proper sources are cited and not link spam claiming to sell this as some sort of service. FSR-Rene (talk) 20:02, 13 February 2026 (UTC)Reply

Also, third parties are not "optimizing" platforms rather, they're consulting/advising people on how to structure web content to align with platform (mainly LLM Artifacts or the Google's search overview, which is also called AIO funny enough. Standards and ranking systems. Engineers developing the LLMs / reasoning models are ultimately the ones doing the "optimizing," but you have to clearly define what you are optimizing for.

For example, the way we wrote our standards, we're primarily optimizing for factual integrity, which can require additional API calls and web searches to verify information. When this process is repeated across many users and queries, the system may later optimize further for breadth of eligibility, while still maintaining those factual integrity constraints. I.e. were ultimately optimizing for factual integrity at the lowest compute cost across multiple users. That is specific to our use case, and perhaps with many other AI models, but it is not the ONLY way to "optimize" an AI system, if that makes sense. FSR-Rene (talk) 20:32, 13 February 2026 (UTC)Reply

Your point of view is making sense to me now. I appreciate it. Sounds like we both think this is TOOSOON. I will note that the article itself, as of now, is not about your standards but is about whatever "artificial intelligence optimization" is overall. The lack of consensus around what it is seems to be murky without a clear consensus. Dflovett (talk) 16:37, 17 February 2026 (UTC)Reply