Google's AI is still broken

[Colin_T]

I was checking something and got the following response from Google's AI. It says Lake Tanganyika has a GH of 814dGH (about 14000ppm). And it gets as low as 711dGH (12,000ppm). Now that is really really hard water.

 

[Tacocat]

GH and KH are so high that the rocks have water in them.

 

[GaryE]

The errors in AI responses are quite spectacular, the way they used to be in translation programs. It's like having a poorly informed person doing your thinking for you.

 

[TwoTankAmin]

Here is the trick to using the Google AI in the search. I am not registered on Google so that is the only place I see the Google AI. The one good thing about it there is at the end of a lot of what it says has at the end of it a small gray circle which what looks like a d link icon inside it. If you click on that it takes you you the site from which if got the information it displayed. Also to the right at the top is usually a list with several sites used for the information the AI offered.

I find the AI answer most helpful when I already know the answer to what I asked and might be able to copy paste what the AI offered and not have to type it all myself from scratch.

I am personally distrustful of AI in general when it is asked to provide facts. My issue is I do not believe AI can provide "opinion" type answers. So I asked Google, how does an AI know what is true and what is not" Here is the answer it gave.

An AI does not know or understand "truth" in the way humans do. Instead, it uses a series of statistical probabilities based on its vast training data to predict and generate a plausible-sounding response. This process is why AIs can sometimes generate incorrect or fabricated information, a phenomenon known as "hallucination".

I have used Microsoft Windows since version 3.0. I resisted moving from & to 10 for a long time. I even tried Apple for a year and disliked. I was eventually force to get 10 so I could continue using TurboTax. I have never had an issue using a windows version for some time after a newer one was released. I had a PC expert set up mu win 10 PC as he disabled a lot of it I did not want.

I am now faced with having to get 11 in order to use TT to file next April. Wen I read about all the AI it will have I started to research how I could turn it off.delete some of it so it could not work. Microsoft 10 is the most invasive software ever and 11 will be worse.

I have never downloaded an app from an app store, I have a local account on WIn. 10. I will likely need a new PC to handle 11. My current plan is to keep using 1- for as long as I can and to get a 2nd machine work 11 which I will only use for TT until 10 stops working in most places.

I do not trust AI at this time. AI is computer software. It is supposed to code faster/better than humans. I wonder if the recent outage at AWS was due to a faulty AI code? If Google is correct that AIs cannot determine truth, how can we trust it?

In college I tokk a number of philosophy courses. One dealt with defining "Justice." We had basically one text we worked with. Plato's Republic. In order to come to a definition of justice, "Plato's "The Republic" typically has around 480 pages, depending on the edition and formatting." Ask an AI to define what justice is and see what it answers.

 

[WhistlingBadger]

I will only use AI for tedious tasks. If AI could figure out how to sweep my floor and do my dishes, I'd be all over that. I'm an elementary school teacher, and I read a lot, and sometimes it's hard for me to create writing tasks that are on a ten-year-old's reading level. So on a recent project I asked ChatGPT to reword my essay to a fourth grade reading level. It did so admirably. But the thoughts and original words were mine; AI was basically just functioning as a thesaurus.

I get really creeped out by the use of AI for the extremely human pursuits of friendship, romance, forming opinions, and creating. It's amazing how quickly any new technology tends to transition from servant to master. (How many of us, as kids, had our family's evening schedule dictated to us by the television?) AI seems to be following the pattern.

 

[TwoTankAmin]

My parents were never big on TV. But I was. Int the early years of B&W TV there was a dearth pf programming. So they played a lot of the movies from the 30s on. As a result, I have seen most of the movies with folks like Kate Hepburn, James Cagney, Cary Grant etc. I would bet I have seen over 300 movies from the early days of TV.

They got me interested in the old films and I have seen a lot of silents as well. I am a big fan of Charlie Chaplin as well as the Marx Brothers. I have seen the very first vampire movie, "F.W. Murnau’s Nosferatu (1922) remains the ultimate vampire horror film — a silent-era masterpiece that continues to haunt modern cinema." I have seen silent films featuring Laurel and Hardy alone before they became partners.

If any reading here want the names of some great older films, I can give you a great list.

​

 

[JuiceBox52]

I find AI to be useful in consolidating information. Start by asking it to retrieve all articles on a particular subject - but then I must skim through them myself, and sometimes the articles it retrieves don’t exist… it hallucinates their existence. Outside of that, which is not asking it for facts, I find it to be unreliable