Google still controls about 90% of the global search market. For over two decades, users have relied on traditional search engines, inputting keywords to uncover webpages deemed relevant by complex algorithms. While this method offers a vast array of information, the quality and accuracy can sometimes be hit or miss.
Now the rise of conversational AI, powered by Large Language Models (LLMs), is rapidly transforming how we search, but will it replace traditional search engines?
Why conversational AI search is becoming more popular
With the increasing use of conversational AI, the way we search is changing at a rapid pace. The appeal of an AI model like ChatGPT in place of a search engine is that you can use natural conversation to ask it questions, without needing to think of terms or keywords. If there’s one thing humanity loves doing, it’s switching off our brains and using technology to help us think less – a dangerous road to travel down.
Which is better right now?
Traditional search engines have had decades to learn from our search queries and understand the context behind certain terms. But AI is learning quickly.
From my experience, conversational AI models like ChatGPT and Microsoft Copilot are great for getting an overview of information and limiting the ‘paralysis of choice’ we may experience from so many search results. However, they are not great at providing accurate information. Many times they have provided information which is either misleading or just an outright lie, and it is so important to ask the model you are using where it has drawn the information from, so you can do a quick check of the source material.
It’s possible to use conversational AI to begin exploring a subject, getting an overview of topics before taking to traditional search engines to explore those topics in more depth. With Google’s E-E-A-T guidelines placing emphasis on credibility and authority, what are conversational AI models doing to uphold this too?
As some users of ChatGPT Search have already observed, most of the time when we use search engines, we’re only using a few words, which is something that Google’s algorithms have become better and better at understanding. So when trying to use ChatGPT Search, it could prove frustrating when using short queries.
How can you optimise your website for conversational AI models like ChatGPT Search?
This field is still developing, but there is generally a consensus that content should aim to answer specific questions instead of relying on keywords, and content must answer those queries comprehensively. It’s already best practice to research the subject you’re writing about in terms of what questions people would like answered, so writers should continue to use this approach.
It’s also going to become more and more important to include links to reliable sources of information, as well as using expert quotes and statistics to add authority.
When conducting keyword research, SEO content writers should also take note of queries that trigger AI-generated overviews. Study AI search results to see trends and preferred answer formats, such as paragraphs, lists, videos or tables. Then align your content with these preferences.
AI also loves the over-use of bullet points and headings, so it will be interesting to see how content creators strike a balance between detailed, quality information and presenting that information in the simplest way possible.
Since ChatGPT Search uses Bing’s index, it’s also a good idea to use Bing Webmaster Tools to check your pages are indexed on Bing and address any gaps.
The future of AI in search
It seems that a hybrid approach is currently the most efficient way of using these technologies and we’ve already started to see search engines adopt this strategy.
Find out more about ChatGPT and how AI is disrupting search engines.
OpenAI’s ChatGPT has now integrated SearchGPT into conversations for paid users, with the feature being rolled out for free, enterprise and education users soon. This will rival other tools like Microsoft Copilot, which can already draw from real-time internet access in their conversation models.
In the meantime, Meta is apparently working on its own AI search solution, so this approach won’t be going away anytime soon.
Conclusion
Optimising content for these AI models continues to be an emerging field and one that we are watching closely. While it is impossible to police the internet, more regulations must be introduced to prevent the spread of misinformation, especially since more and more inaccurate content produced using AI is circulating the internet, and this is in turn being used to feed these AI models.
