AI Assistant
Notifications
Clear all

AI "helper"

8 Posts
7 Users
9 Reactions
1,622 Views
MightyMoe
(@mightymoe)
Posts: 10540
Member
Topic starter
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

I don't know what goes on with the AI when you use a search engine. I simply wanted to check gas prices for a town I'm going to be driving through. The gas there is always so much cheaper than my town. Anyway, I ask the default search engine what the gas prices there are. The blue text box said gas price is $3.45. 

WHAT??

So, I tried again; same result. I'm thinking the AI is wrong so I clicked on some other search results and the real number is $2.07, premium is $2.49 and it's one of my "approved" venders. 

How is the AI so wrong? I've noticed the same thing before and I've come to the conclusion it's the same AI that builds county GIS. 


 
Posted : December 6, 2025 12:20 pm
mathteacher
(@mathteacher)
Posts: 2244
Member
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

If you're using Google, then it often gives wrong answers and corrects itself when you "dive deeper." Often contradictory info between first answer and dive deeper answer. Also it's better when you ask very specific questions including dates or specific locations.

ChatGPT is far superior, but it ain't perfect either.

I have no experience beyond those two, but those two are like people in a way. Some can be trusted, others can't.

 

 

 

 


 
Posted : December 6, 2025 3:04 pm
MightyMoe
(@mightymoe)
Posts: 10540
Member
Topic starter
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

I don't think AI is intelligent. The mistakes that pop up when it's activated leave me wondering if it's more of a troll generator than anything else. 


 
Posted : December 7, 2025 2:24 pm
jimcox
(@jimcox)
Posts: 2102
Member
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

What with AI 'Slop' and 'Hallucination' (gotta love that euphamism) its just another example of the enshitification of the modern world


 
Posted : December 7, 2025 4:00 pm
2
not-my-real-name
(@not-my-real-name)
Posts: 1216
Member
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

When I compose an email AI asks if I want help, so I dismiss. Then when I am finished composing AI asks if I want help 'polishing' my writing, so I dismiss. Next time I try to write, AI is there asking the same questions. So, if it is really intelligent, then why can't it take a hint?


Historic boundaries and conservation efforts.

 
Posted : December 7, 2025 6:12 pm
3

Landbutcher464MHz
(@landbutcher464mhz)
Posts: 222
Member
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

@not-my-real-name "So, if it is really intelligent, then why can't it take a hint?"

Exactly. It is basically intruding into my email and just repeating stuff that is already in the email. Like it thinks I don't know how to read. I am looking for the OFF button on that sucker.


 
Posted : December 7, 2025 7:35 pm
2
murphy
(@murphy)
Posts: 948
Member
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

AI never forgets so be nice to AI now or you may regret it later...

AI works great as long as you're specific and redundant in your command language. It's learning fast, but right now there are lots of ways a large language model can mistake unrelated or false information for truth. It's scary how good AI is already which makes me worried when I here people seemingly delight in the "mistakes" AI makes. AI is going to get much, much better. It's already a better proof reader than anyone in your office. Drop a PDF or Word doc in there and give it specific instructions such as: Proof read this proposal for land surveying services for grammar, clarity, conciseness, and general professionalism using NC jurisprudence and general contract law as guidance.

Gemini has been intensely trained for medical purposes. Obviously, you shouldn't blindly follow medical advice from AI, but I'd say the same about blindly following medical advice from many GPs. If your doctor isn't using a medically trained AI in conjunction with their own research, they're not providing you the best service available. Again, don't view AI as a replacement for anything, view it as a check or second opinion.

 


 
Posted : December 8, 2025 7:44 am
2
jaccen
(@jaccen)
Posts: 458
Member
Translate
English
Spanish
French
German
Italian
Portuguese
Russian
Chinese
Japanese
Korean
Arabic
Hindi
Dutch
Polish
Turkish
Vietnamese
Thai
Swedish
Danish
Finnish
Norwegian
Czech
Hungarian
Romanian
Greek
Hebrew
Indonesian
Malay
Ukrainian
Bulgarian
Croatian
Slovak
Slovenian
Serbian
Lithuanian
Latvian
Estonian
 

"AI" is too broad a term. I, personally, use some of the LLM's for helping "story boarding" a process, for help with initial coding of small projects, and as a research tool via data analysis on a cloud share that I specify and nothing else (ie. it does not go out on the internet--it only uses my tailored data sources). ML has been doing pretty good work in surveying already (ie. most of the point cloud applications and their auto-classify operations).

 

We recently used it to help program an application that allows us view 360 images taken with a GoPro Max 360 as though it was Google Streetview. It allows us to keep the data private (ie. limit liability for license plates, faces, etc. as the data is not shared publicly), use it on private sites, tailor it to our exact needs while not paying another software subscription. Admittedly, we used the pro version of one of the LLM providers, but 1 month at $20 is much easier to swallow then whatever Autodesk or the like charge yearly. You, as of now, can cancel whenever you like. I recently cancelled one for another. 

 

It is not, IMO, useful for engineering or surveying calcs. It does have its place as a research tool and coding aid for the coding limited (aka "fat fingered" me).

 

There's no way I would allow it to scan my email, company docs, or anything that I would not feel comfortable publicly sharing on an open FTP to the world.


 
Posted : December 12, 2025 5:11 pm