When AI Overtakes SEO

There's a lot of discussion emerging about the impact of AI on the future of search, and how that's likely to impact marketing, particularly SEO.

There is no question in my mind that search will be replaced by AI in the near future--it's easier to use, produces more relevant results, and synthesizes information in a way search can't.

Personally I'm guessing 18 months before we see significant drops in search traffic. There is also no question in my mind that marketers will rapidly pivot from SEO to AI optimization, and start figuring out ways to game the system to influence AI results and elevate their own messages. So where is this likely to balance out?

Turns out, this is a great question for AI. So I had a little chat with GPT4 to game this out. Here's what I learned:

Q: Given the direction of AI, it seems clear LLMs will replace search for most users. As AI becomes more popular, businesses will start trying to optimize their content for AI, much like they've done with search engine optimization. What should marketers be doing to make sure their products stand out to AI systems?"

A: Here are some things marketers should prepare for:

• Understanding AI Algorithms: To optimize for AI, marketers will first need to understand how AI algorithms work. While these systems are complex and often proprietary, some general knowledge can still be gleaned. For instance, it's important to know that these systems prioritize accuracy and relevance in providing information, which often involves analyzing the context and intent behind user queries.
• High-Quality, Contextual Content: Given that AI prioritizes accurate and relevant responses, the need for high-quality, well-researched, and contextually accurate content will increase. This means that keyword stuffing or superficial content strategies aimed at "gaming" the system will likely be less effective. Instead, marketers should focus on creating value-driven content that thoroughly answers potential questions from users.
• Structured Data and Metadata: Making sure that data is well-structured and metadata is accurately labeled can help AI systems better understand the content, thus making it easier for these systems to reference and deliver the information when relevant.
• Transparency and Trustworthiness: As the importance of transparency and fairness in AI grows, organizations that ensure their methods are transparent and their content is reliable and unbiased will be more likely to gain favor not only with AI systems, but also with users who value these characteristics.
• Interactivity and Engagement: AI systems are likely to increasingly incorporate engagement metrics into their algorithms. Therefore, creating content that encourages user interaction and engagement can potentially improve visibility.
• Ethical Considerations and Regulatory Compliance: Ethical considerations will become more significant, and regulations to ensure fairness in AI are likely to be introduced. Companies will need to navigate these changing landscapes carefully, maintaining up-to-date understanding of any legal and regulatory requirements pertaining to AI.

It's also worth noting that the process of AI optimization may differ from traditional SEO in that it may be less about manipulating a system to gain visibility and more about genuinely providing value to users in a way that aligns with how the AI evaluates and delivers information. This makes it somewhat harder to "game", but it also levels the playing field to some extent, allowing companies that genuinely prioritize delivering value to users to shine.

Q: I'm concerned about the idea of using engagement as a measure of value. These days, a lot of engagement happens on social networks, which are designed to keep people on the site for as long as possible to boost ad revenue. How can AI tell the difference between genuine engagement and the kind that's been engineered by social networks for their own purposes?

A:Balancing genuine engagement with content quality and relevance is a significant challenge, especially when AIs try to sift through the noise and deliver trustworthy and accurate information. Here are a few strategies that may help address this challenge:

• Multidimensional Analysis of Engagement: AI can take into account more than just likes, shares, or comments. It can also consider the context and content of these interactions, differentiating between shallow and deep engagement. For example, an in-depth conversation in a comment section might be weighted more heavily than a simple emoji reaction.
• User Behavior Analysis: Understanding the behavior of users who interact with the content can also provide useful insights. For example, a user who spends significant time reading a post or article, or who returns to it multiple times, is likely more genuinely engaged than a user who just skims it quickly.
• Content Quality Assessment: AI systems can evaluate the quality of the content itself, looking at factors such as the depth of information provided, the accuracy of the information (based on its agreement with other reputable sources), the language and tone used, etc.
• Source Reputation: The reputation of the source of the content is also a useful factor to consider. Content from sources known to be reliable and accurate might be prioritized over content from less trustworthy sources, regardless of the level of engagement.
• Checking for Engagement Manipulation: As AI technology improves, it could become better at detecting signs of engagement manipulation, such as sudden spikes in engagement that don't match usual patterns, or the presence of bot activity.
• Factoring in User Preferences and Feedback: AI systems can also learn from the preferences and feedback of individual users, so that over time, they get better at delivering the kind of content that each user finds valuable and engaging.

However, it's important to acknowledge that these strategies are not foolproof, and this is a complex, ongoing challenge. As AI technologies continue to evolve, they will need to keep adapting to meet these challenges and deliver the best possible results for users.

Q: The stakes for business are obviously quite high, and they have the resources and creativity to try and game systems in whatever ways they can to influence search results. Clearly they will do the same with AI. It's easy to say that AI will bend the curve toward more transparent and trustworthy information, but for businesses with vast resources that don't have the best competitive products available, generating trustworthy and transparent content will take a back seat to strategies for influencing AI output. What tactics are likely to emerge for those companies to beat AI and influence the elevation of their own messages?"

A:Certainly, businesses with significant resources will continue to innovate ways to optimize their visibility in AI outputs, just as they have with search engine results. The exact tactics will depend on the specific mechanisms and algorithms used by the AI systems, which could vary considerably. However, here are some possibilities:

• Semantic SEO: Instead of optimizing content for specific keywords as in traditional SEO, businesses might focus on optimizing their content for the way AI systems interpret and understand language, using techniques like natural language processing. This could involve using more natural and conversational language, addressing related queries and subtopics in depth, and providing clear and concise answers to common questions.
• AI-Generated Content: Companies could use AI to generate their own content, ensuring it's tailored to the preferences of the AI systems they're targeting. This content could be dynamically generated to match trending topics or frequently asked questions, ensuring it's always relevant.
• User Behavior Analysis: Similar to how marketers use A/B testing to optimize click-through rates in email marketing, they could use user behavior analysis to understand how users interact with different types of content and optimize their content accordingly.
• Influencing the Training Data: If a company knows which datasets an AI has been trained on, it might try to influence those datasets. For example, it might try to get its own content included in the training data, or even sponsor the creation of new datasets that reflect its preferred narratives or perspectives.
• Partnerships with AI Companies: Some businesses might seek partnerships with AI companies or platforms, in which they provide funding or resources in exchange for preferred visibility in AI outputs.
• Manipulating User Engagement: As with search engines, there might be attempts to artificially inflate user engagement metrics, in order to make content appear more popular or relevant than it really is.

It's important to note that many of these tactics could potentially undermine the trustworthiness and neutrality of AI systems. For this reason, it's crucial that AI developers take steps to safeguard against such manipulations, and that regulators consider how to ensure fairness and prevent misuse. The AI industry will need to continue advancing and refining its techniques to detect and mitigate such manipulation tactics.