Dreadful AI Result

in my humble opinion, AI is the antithesis of intelligent. AI chatbots may be able to scan the internet for gobs of information much faster than I can but, as you have pointed out, they cannot ensure that the data is correct.

I've had better luck with ChaptGPT, acraven, although it did give me an itinerary that included visiting the Louvre on Tuesday when it is closed. But overall I've been pretty impressed with some of the results I've gotten.

Very interesting! That jibes with my limited experience: I would rather do the work myself and know it is correct.

I would rather do the work myself and know it is correct.

Mama was right. "If you want it done right, do it yourself".

I have lots of friends lately who have been extolling the virtues of various AI searches.

I’m holding out firmly in the camp of not using it. I would just rather look it up myself.

But I was forced to listen to someone recently as they consulted some AI site about me - specifically. And the site provided some regurgitated sentences from my business site and paraphrased a client review of my legal services from AVVO.

That’s it - omitting my travel website, my appearance on a national radio show, and lots of other key details.

In addition, two “facts” it provided about my business were wrong.

The reality is the word artificial.

I prefer the real thing. Sorry Coca Cola.

AI= Automated Idiocy

And for a warm n fuzzy feeling, the geeks want these things to pilot driverless cars.

AI= Automated Idiocy

I will be using this phrase a lot.

LLM's made their debut in the public arena with the release of chat GPT in November of 2022. that was just over 2 years ago. while I am not a computer scientist I am involved with exploring AI professionally in my field. the progress has moved at a clip never seen before. where it's headed no one knows. good or bad I plan on learning as much as I can otherwise I risk becoming 'irrelevant'.

And for a warm n fuzzy feeling, the geeks want these things to pilot driverless cars...

...and the planes you will be flying in (and Air Traffic Control, and more).

I’ll take the word of a Sentient Being
Over AI any day of the week!

I would rather do the work myself and know it is correct.

Amen

And waste a lot less time, too.

Thanks for sharing your experience, acraven so I don't have to go through that !

You are a travel advisor with a client who loves museums. The client will spend at least 1 week in Glasgow during May 2025. He has asked for a list of the opening days and hours for all the museums in Glasgow.

To be slightly fairer to AI here (and I personally have little confidence or interest in AI), my first two questions to the Client would be-
a- Are you aware of the two public holidays in the month of May?
b- What do you define as Glasgow? (that may well broaden the client's knowledge level) , with a subsidiary question about the Client's interests. There is no point in sending the client to Art Museums if 'he' has no interest in art, but is a big social historian. Or to a police museum.
c- How is the client intending to travel round the city?

To me this question is too broad. The narrative given above shows that refining the question gives better answers.

The definition of Glasgow is important- in my eyes it is not just the City Centre but Greater Glasgow up to the various surrounding County Borders- what I would broadly define as the area covered by the former 'Blue Trains'- the suburban rail service.

On this forum I see the same few museums given on here all the time which in no way covers the full range of museums in the city.

As just one example - Helensburgh- all that is ever mentioned on this forum is the Hill House. I never see the Scottish Submarine Centre or the Outdoor Museum mentioned. That is before other things in the town like the Hermitage Park and the renowned Cherry Blossom around the streets of the town (could still be there in early May).

As another question- if the Client has a niche interest in the Police Museum, then perhaps it would be good to give him a list of the few surviving Police Boxes in the City. They may not be museums but are part of that history.

I’ll take the word of a Sentient Being Over AI any day of the week!

Next week AI may be a sentient being.
https://archive.org/details/tvtunes_3958

isn31c, that search wasn't an academic project. It was something I used for the purpose of (attempting to) get information I need for an upcoming trip. I wasn't looking for value judgements; I already have a list of all the museums in Glasgow city potentially of interest to me or my travel mate. I'd never use AI to vet that sort of information; I use sources like guidebooks, tips from this forum and tourist-office websites to identify sights of interest. I just wanted to save myself some effort in updating museum hours, because my most recent visit to Scotland was in 2019. I wouldn't have objected to inclusion of museums located outside Glasgow proper; indeed, I have no clue about the city limits of Glasgow. What was ludicrous was that initial list of only four museums when I had asked for "all". Perhaps the presence of the police museum and football museum are the result of my use of the pronoun "he". I speculate that the bot thinks such places to be of more interest to a male visitor (I'll note here that I'm female) than the world-class Burrell Collection. That sort of value judgement is what I was trying to avoid by asking for "all".

I'm aware of the two bank holidays in May. I haven't yet seen info about museum closures on those days, but I haven't manually checked all the museum websites yet. I would have expected CoPilot to include information about holiday closures in the month of May 2025, given that my query specified a visit during that month. I don't think the user should have to tell the bot about holidays in the month of interest.

I do wonder--given the egregious inclusion of a museum that has been closed for renovation for some time--whether the bot is using an outdated database. If that's not it, there's a huge reading-comprehension problem; the hours of operation are laid out very clearly on the websites of the museums I checked manually.

www . 3blue1brown . com / lessons / gpt

The above is a link to one of the best websites that explains how AI actually works. If you are curious, it is worth the time. Its origins are akin to autocomplete (or autocomplete on steroids)-- that is to say it just predicts the next word and then word after that and so on. When it starts to construct an answer, it doesn't "know" where or how what we see as a complete thought ends.

In another lecture it discusses a recent paper on the topic-- I am paraphrasing-- where does AI/Machine Learning actually store the fact of "Michael Jordan plays basketball" and if so, where is it stored. The answer in the paper is "we are not sure where it stores it-- but probably here-- and nor are we sure it actually stores it as a "'fact'". Inspires confidence, no? "Facts" are not a machine learning "thing".

So it is 100% for sure without a doubt unreliable-- but often right-- and programmed to bllsht with élan and utter confidence. A good resource explaining this is from U of Wash professor Carl Bergstrom.

Here is link on the topic: https://theb*llsh*tmachines.com. [Replace the * with the right vowels.]

Anyway, I 100% agree with you on not trusting AI for travel advice. What maybe we should do instead is clone Rick Steves. Ah, human cloning and travel advice!! I think I'm on to something. Let me write that one down.....