OpenAI, up to this point, has shown a willingness to outcompete some of the very companies that rely on their API to function, most famously with the release of GPTs, which had a quite severe impact on many "AI startups" [0]. In that way, they remind me of Apple with Sherlock way back in the early 2000s.

Despite this, I was doubtful that they'd go so far as to release a full-on search product due to their relationship with MSFT and reliance on Azure credits. I am happy to admit that I have long stopped any attempt to properly understand how OpenAI's corporate governance and company structure work, so I have a hard time following where this falls under and who would decide on this release, as well as how they interact with the part of OpenAI cooperating with MSFT and the Bing team, but I still have a hard time seeing how releasing a clear Bing competitor wouldn't cause some trouble for their entire suite of products and maybe even hinder future expansion by limiting the resources they can rely upon.

I am also interested in how this will impact search in https://chatgpt.com/, which, like everything in that product, has been inconsistent to a maddening extent. Started out barely usable, failing consistently, then got reliable whilst retaining the ability to search through multiple sites and handle more than one request in a row, then lost most of those capabilities and now barely works anymore, only looking at an incredibly limited, often barely fitting selection of results, whilst also needing to be manually invoked by asking for a search, rather than before when that was done automatically whenever it seemed sensible.

Like so many changes, e.g. the subjective reduction in GPT-4's abilities over time whilst retaining the model's name (not to mention the regressions they publicized in the name of efficiency, like the "turbo" variants), this is certainly done to reduce costs to the point of finally becoming financially viable at the $20,- price they charge for ChatGPT+. I might be in the minority, but I will continue to scream from the rooftops that I am more than willing to pay far more for a consistent, guaranteed, high-end LLM with web access (which sadly excludes Anthropic's efforts).

[0] I still dislike using that term for companies solely relying on third-party API's, a frontend and database solution, especially since I also detest calling LLM's "AI", but it's what this crop of companies have been termed and how they collected bucket loads of VC.