The AI search engine arms race: OpenAi are testing a new search engine, “SearchGPT”

OpenAI announces it is creating a new search engine 

The prototype was launched last Thursday with a small group of users – reportedly including around 10,000 people and publishers. The invention of this new search engine could potentially rival Google’s dominance of online search and have huge consequences for the SEO industry, if users adopt it. 

OpenAI stated the temporary prototype (SearchGPT), will combine the company’s existing AI models including ChatGPT and the ability to search the internet.

The new search engine summarises information provided from websites then creates short descriptions with attribution links, as opposed to linking directly to a website.

In December, the NY Times filed a lawsuit against OpenAI and Microsoft accusing of the “unlawful use of millions of copyrighted articles” being used to train AI models. In response, OpenAI filed a motion to dismiss parts of the complaint, stating users did not use ChatGPT or other OpenAI products to substitute a NY Times subscription. According to Forbes, the lawsuit is still ongoing.

Our SEO team have shared their thoughts on what this could mean 

Ash Graham, SEO Strategy Manager at 26PMX shares his insight to this news. 

In the past, Google gained search engine market share because it returned the highest quality results and provided the best user experience with fewest ads. To match this, SearchGPT has 2 barriers it would need to overcome to gain a large market share. 

Barrier 1: Can Search GPT return high quality and accurate results? 

OpenAI will have similar challenges to Google AI Overviews which initially launched on 64% of search queries, and then reduced down to 8.7% due to the quality of results. See the infamous "drink your own urine to pass kidney stones" response below.

SearchGPT has  side stepped one of the issues AI Overviews through showing clear sources and attribution links and partnering with large reputable publishers, such as The Athletic. This will increase the amount of trust users have in the answer provided and the content quality.

If they can consistently return high quality results, which answer user queries through their LLM (Large Language Model), then SearchGPT has a higher chance of succeeding. However, Google currently has the advantage in terms of data it can use to train and improve AI Overviews through Chrome (3.45 billion), YouTube (2.49 billion), Gmail (1.8 billion) and Maps (1 billion) in comparison to just the users of Chat GPT (180.5 million).

It will be interesting to see how this pans out as the recent demo failed on this part. 

Barrier 2: Is mass market reach and market share possible on a paid model?

Google is free, and users are not accustomed to paying for use of their search engine (due to selling advertising). If SearchGPT requires a ChatGPT membership starting at $20 a month, this will limit their market share. 

As the CEO Sam Altman is very much against ads, it will limit their potential market to paying ChatGPT users. With the US being estimated at 3.9 million paying subscribers vs 8.5 billion daily google searches.

What should businesses do?

1. Join the waitlist to experience SearchGPT and to understand how this changes the search landscape
2. Keep an eye out for when it gets rolled out to a wider audience and any information on adoption
3. Start evaluating how much coverage AI Overviews has by using a tool such as 'Ziptie.dev' (if you target the US market). This will reveal the main competitors and you can reverse engineer what is working to get into AI Overviews.

Need help navigating your SEO strategy? Contact 26PMX. 

Read more about our services here: https://26pmx.com/services/