OpenAI insists it's not launching a search engine nor GPT-5 on Monday OpenAI, the maker of many chatbots and taker of much Microsoft money, denies it's planning to unveil a web search engine on Monday. "Despite reports, OpenAI isn’t launching a search product or GPT-5 on Monday," a spokesperson told The Register in the past few hours regarding a product update coming early next week. CEO Sam …
If AI search aspires to rival web search, it needs to reveal discoveries presenting absolutely fabulous fabless opportunities for lend-lease sale and ransom payment subscription charging to leading novel market pioneers into the building of futures and derivative support ventures .... for AIdDevelopments of Advanced IntelAIgent Design in both an attractive and addictively rewarding alien space and human place ..... Live Operational Virtual Environment.
Tell us then that is not a Great Game Changer and bask in the drowning tsunami glory of richly deserved waves of peer derision and pity.
AI happens to be an expensive habit and it looks likely to become more so as data owners circle the wagons to prevent permissionless training and strike licensing deals.
There are millions upon millions of competent literate people who read far fewer books and slurp far less social media garbage, and yet are far smarter than ChatGPT in terms of communications and generalization ability.
Doing more with less would bring the cost down at lot, and it might even improve the utility. The stress of shrinking resources has often been a driver of human progress.
I think we are now entering the phase of consumer AI where progress is asymptotic. We have seen all the party tricks but expectations are now far higher than the technology will be able to deliver. We last saw this with self driving cars of course... Oooh all that investment; is a bubble about to burst?
Wiretap, Hi. Nice to hear from you, to hear from you, nice.
And I have a question relating to your post entitled .... LLM progress is now asymptotic
Can the same be truthfully said of AI progress in Secret IntelAIgent Services for deployment and employment in secure private hands of the capture of hearts and minds serially troubled and in the raptures naturally associated with the advent and indisputable evidence of extremely disturbing discoveries?
Methinks betting on that being true would have you shirtless in next to no time at all ...... but to quote/paraphrase Mandy Rice-Davies ....... They/He would say that, wouldn't they/he. :-)
Poe's Law Rules Apply . And it's a funny old strange world, is it not. So full of so many truly unexpected surprises.
OK, so I can be a bit slow. It never occurred to me before that my expectation of what a search engine is and does might differ from that of the organisations that actually run them.
When I use an online search engine I am searching for web pages, or possibly entire web sites, containing the keywords I specified. How naive. It turns out that the search engine believes that I am seeking disembodied factoids shorn of context. No wonder I am increasing disappointed and frustrated.
-A.
"When I use an online search engine I am searching for web pages, or possibly entire web sites, containing the keywords I specified."
Quite right, too, and don't bugger about with the results based on what else you think I might be interested in. Just give me what I ask for, and I will decide if it's useful or not.
But there is another kind of query, where you are looking for a simple answer to a simple question, where it is useful to have the search engine simply give you the answer (with appropriate links to references, of course). I find that quite handy for simple factoids that my aging brain once knew, but can't quite haul up from deep storage right this second. Would I trust those answers to be correct without verification? No, not for something important. Is it handy to get a quick summary? Yes.
That is arguably more the function of an encyclopedia, or information engine, than a search engine, and I should imagine that sites making money from pages of facts are not happy, but it is useful to the user. I also imagine that, in the right circumstances, a quick summary of other things - or perhaps several related things - might be useful, too, and an LLM would be great for that IF it linked to its sources and didn't hallucinate. But since it doesn't and does, respectively, it's not there yet and may never be there.
chatGPT isn't supposed to be a search engine, and it always amazes me that people try to use it as one. It could be an answer or information engine, if it were reliable - maybe that's what openAI have in mind for their announcement?
"Keyword search works well for document retrieval and has become quite effective"
Please tell me where. Everything I've seen gets progressively less effective. Is sheer quantity of hits with some of the keywords or some words which look a bit like keywords a measure of effectiveness? What happened to the effectiveness of "AND", "NOT" etc that worked in document retrieval systems of the '80s but not on the moder web?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
