I also read the Google boys came out of retirement (since 2019) recently to be part of a meeting discussing OpenAI and how it is a potential competitor to their search engine.
Exactly... yet you don't consider Google [advanced] Search an AI, do you? They do the same thing (though slightly differently) in scraping the same Web and put it in the database (again, differently, utilizing different distributed file systems to optimize retrieval for given purposes - GoogleFS/GFS / Colossus was also a model for Apache Hadoop HDFS designed for Big Data store on clusters of commodity hardware, though now is common on dedicated HCI (Hyperconverged Infrastructure) hardware such as HP, Dell, Nutanix etc., with various VM hypervisors.
At some point buzzwords "Big Data" and "analytics" became passe, boring and unimpressive and didn't attract as much VC money, so startups started throwing "AI" name into anything and everything that required minimum of "pathfinding algorithm" (see Edsger Dijkstra, 1959) - money is still flowing to that space, so "AI" is still golden.
Watson was built for providing AI services to integrated applications and specific industries such as Health Care, Finance and others.
Exactly, Watson's storage / databases were fed with much more of specialized knowledge, like manuals, transcripts and videos (including 3-D) of medical, financial and legal information because at some point, after failing to make it attractive to general-purpose marketplace, they decided to sell it into these industries and have been quite successful. Doesn't make it AI, but can significantly reduce human assistance, just like general-purpose computers with general/special-purpose software have done before.
Does spreadsheet strike you as AI? Was it "magical" / "wow" when you first saw a demo or used Lotus 1-2-3 or Excel? But it can manipulate numbers and very complex formulas. Chatbots and ChatGPT can manipulate [stored] text and wrap it into coherent content in the manners its programmers think you would like or expect. Just like half a century ago there were different computer languages for different tasks - FORTRAN for math, COBOL for accounting/finance, ALGOL for text manipulation, RPGIII for report generators etc.
OpenAI is geared towards bringing AI to the public for the "good of humanity".
Yeah, sure, fine. Remember Google's original motto "Don't be evil" which became Alphabet's "Do the right thing"? About the same thing, with a slightly different sauce / wrapper, no?
Chatbots are designed to be more verbose than just "here is the link with short description / first paragraph, you are a big boy, decide which one you need and... click / tap." Also (here's the important part!) Open AI right now is not driven by optimizing for ad revenues (i.e., in same place as Google was when it started to compete for search with Altavista, Ask Jeeves, Lycos, Xtreme, Yahoo etc., before Eric Schmidt found ways to monetize it with GoogleAds etc.) - they need to attract people tired of ads and self-serving placements, so they want you to like them for what they are, without having to think about revenue production... for now.
I asked OpenAI to explain the difference between itself and Watson. It said they are both AI platforms with differences emphasis.
No kidding, those chatbots / databases / "knowledge stores" like to brag about being AI, don't they? Did you expect it to say "Nah, I am just overstuffed database with chatbot output"? Ask Google Search if it is an AI? Do you think it will say "Nah, I am just a search engine"?
ML, to scrape and gobble up info - yes. AI - no.
"Alexa, when is the end of the world?"