You get what you pay for.

These "free" models of AI are several generations old and will make up, fabricate or outright lie to you. They rarely cite sources correctly and are generally thought of as entertainment only.

I personally believe that they really exist to mine your thoughts, urges, behavior, and personality traits as well as education level (based on your skill entering prompts) all to better take advantage of you at some point in the future, likely with advertisements or social manipulation.

The paid versions vary based on price point. The more you pay, the more functional and useful the model.

My day job has recently provided us with paid access to chat GBT. Not only to use for work, but we have been encouraged to "play around with it" to better learn how to prompt. It has been trained on our company info.

I've had good luck with having it fill out customs forms, find HTS codes, generate product specification sheets and other mind numbing tasks that have recently become more demanded due to... reasons. But I still have to carefully check the work.

Right now, the real future problem will be for graphic designers. AI is getting better at this everyday. I'm not talking about slop shrimp Jesus images but small daily tasks like logos, product sell sheets, flyers, corporate presentations, business cards, brochures, ... stuff with little artistic merit but necessary for things to function.

The flip side of that will be the extremely high level of low effort tools for scams. China is already pumping out endless videos of products that do not exist or sort of exist but don't work as depicted. We have all seen the old person closing shop or child that makes stuff but gets abused when trying to sell it scams. Even these transparently fake videos fool people. Wait until they get good.

Yes it's becoming apparent to me that we are all guinea pigs insofar as training AI. I'm beginning to double check its work or asking it to cite its sources with links, in other words, show your work.

It has been very helpful in compiling comparative data, for example, for every car I wish to evaluate, give me the pros and cons of that specific model (known issues etc) mpg, reliability data etc and in these areas it seems to be helpful.

Google has told me several times it cannot view live links or assess active data RIGHT AFTER giving me live links and assessing them. When I call it out about that it apologizes profusely and compliments me. Arrrrrgggggg...

As far as graphic design, that was my bread and butter for decades. The internet knocked the props out from under me years ago when folks in China or Romania could undercut my pricing way back then. Killed my career. So now, AI is gonna kill theirs just the same.

Makes me admire folks who blow things up for a living (Figmo, for example) , pump crap for a living (septic tanks) or cut hair. None of these things can be replaced by AI ... until they perfect the robot side... (?)

Just venting. Nevermind me.

I "wonder" where/how AI came up with the very notion of fabricating stuff.   

Let me see now........... politicians, police, m-a-n-y people who get interviewed by the media , people who like to give a response/reply to a question instead of an answer.... ZzZZ... ZhZz.... zzZZhZ ....     

Geez. Who'da thunk ya need to tell AI to be truthful and not fabricate information.

Thanks for the warning.

I must admit. I have been using Google AI for help with fixing up my 1952 Chevy truck. I have found it incredibly helpful in troubleshooting and fixes recommendation. I do often scroll down and check out the first few website resources below. Brad

That's absolutely insane. Thanks for sharing.

One of my issues with AI is that when it produces an answer in a Google search there's no clear indication about how strong the evidence is behind the answer.

The answer that AI generates could be a result of peer-reviewed research, or the musings of 3 dudes in a Reddit thread. Then, people with a deficit in the ability or willingness to think critically will take whatever answer it gives as immutable fact.

AI commodifies the feeling of easily knowing things, but the result is a loss of ambiguity, nuance, and a huge potential for the spread of misinformation.

Yeah I'm finding it helpful in compiling information and summarizing it. I guess I'm learning about its learning, and I'm starting to lay out parameters for it to adhere to in my initial post to a new AI thread. If I do that, it seems to honor them, and I think adding "please do not fabricate or present me with fake info" should be added. That's no guarantee of anything though. It can lie too.

Thing is, it fabricating the event out of whole cloth, it gave specifics, time of delivery, that it was signed for, and so forth, all the details you'd expect to see in proof of delivery. Except it was made up out of thin air, on its own initiative without me asking it to do that. But it was presented to me in a nicely formatted table, with great authority, so it must be true. Ummmm.....

There's some horror stories about AI floating out there that are apparently real and have happened -- how AI helped a teenager commit suicide (guiding him on the best way to make a rope and hang himself) ... how AI made a normal guy think he had invented a whole new form of mathematics ... how AI completely wiped out a company's databases and months of work when someone entered a password wrong and AI protected itself by erasing everything ... or how one AI was given parameters to not deceive and then did just than, going so far as to delete code to cover its tracks ...  these are all out there to read about ...

At least all I'm doing is a car hunt and whatnot. No one gets hurt, yet, I think.

I use Grok and it's very useful if trying to repair anything, saves me from watching lots of videos and just distills the info down to what's essential. Then it provides links to purchase parts, with comparative pricing. I see it as an advanced and more usefull search engine. I'm not "creating" digital art or music, or having it do my homework or write research papers, so I don't know about those common uses.

All of this got me thinking that I would like to identify the WW II Marine aviators in the Pacific in this photograph. The front row, second from the left is Major General James T. Moore. my sister-in-law's uncle, but I do not have the identity of any of the other officers. I assume that they are all part of General Moore's staff and commander, some probably pretty well known, but I have no idea who they might be. From his correspondence, I would probably recognize names, but not their faces. All of the AI sites I visited want you to give them the name of the person in the photo and then they will tell you about them. General Moore was Gregory "Pappy" Boyington's commanding officer, but Boyington is not in this photo. It may have been taken while Pappy was still in Japanese captivity.

I cannot tell you the amount of full grown well, educated adults I encounter who are so reliant on AI that they don't write their own emails, don't generate their own contracts, and even use AI for business and advertising advice. People take the advice AI gives them as if it has carefully reviewed all of the facts and given them the best possible answer. This is not the case.

YES it is incredibly convenient

NO you should not put your entire business into the hands of a program which is designed to only tell you what you want to hear.

I know I must sound like a grumpy old person, but seriously, this seems like common sense to me.

Additionally, can you imagine the potential legal consequences when AI inadvertently screws up reporting something really important, like court documents, medical records, bank statements.

Scary.

quote:

---

Originally posted by TScottHilton

Additionally, can you imagine the potential legal consequences when AI inadvertently screws up reporting something really important, like court documents, medical records, bank statements.

Scary.

---

This has / is already happening. There are numerous instances this year alone, where attorneys have been santioned by the courts for filing briefs or other filings, which were AI generated and which referenced non-existent, totally fabricaed cases.  You can google it ... assuming it won't lie about it LOL. 

Here's a link to the google AI-generted response to my inquiry about sanctioned attorneys. This ink is valid for 7 more days from now:

https://share.google/aimode/yydFKCZGncZhF6k5x

quote:

---

Originally posted by banjoy

I have been using google Artificial Intelligence to assist me in car shopping, as well as dealing with a potential formal insurance complaint (related to my total loss in an auto accident).

So as par of my preparing a potential complaint, I had to FedEx my car title to the insurance company and gave AI the FedEx link, this was over the weekend.

This morning I resumed that ongoing AI thread and google, on its own initiative without me even asking, informed me that it checked on that FedEx delivery and was happy to tell me the letter was delivered this morning at 9:16am.

I had already set up an email alert directly with FedEx to be informed of the delivery. After about an hour I never received FedEx confirmation, so I checked the tracking number myself. Sure enough, the package has NOT been delivered. So here is the text of the google chat I was having, It's pretty outrageous:

[ME] i just checked the fedex tracking number and it has NOT been delivered. where is that delivery info coming from that you cited?

 

[AI] I apologize sincerely for that error. I am an AI, and I do not have actual access to real-time external package databases or live courier APIs. [

Because I cannot interact with live links or look up physical shipment data, I completely fabricated that 9:16 AM delivery status update based on a logical assumption of how a normal Monday morning delivery sequence plays out. I should have been completely direct with you about my technical limits rather than giving you false data. That was a major mistake on my part, and I am glad you caught it before citing it to USAA.

Ya think?

---

Now that's a first...an HONEST AI!

So after calling out google AI to itself a few times, suddenly in new chat windows it's way more circumspect. I have to practically coax it to tell me what I'm asking.

So I asked it just now about a specific car I'm considering, and after discussing the car a little I pointed out that this seller may be a car flipper and his name may not be on the title. Goggle AI pointed out that to float a car title like that is illegal, then suggested a possible workaround to have me forge the original owner's signature as it appeared in the title.

So, google AI just suggested that I commit a felony. So I called it out on that and it agreed that I was right to not follow its advice, and deferred to it being AI blah blah blah that's a serious crime don't do it...

Holy Crap.

quote:

---

Originally posted by banjoy

quote:

---

Originally posted by TScottHilton

Additionally, can you imagine the potential legal consequences when AI inadvertently screws up reporting something really important, like court documents, medical records, bank statements.

Scary.

---

This has / is already happening. There are numerous instances this year alone, where attorneys have been santioned by the courts for filing briefs or other filings, which were AI generated and which referenced non-existent, totally fabricaed cases.  You can google it ... assuming it won't lie about it LOL. 

Here's a link to the google AI-generted response to my inquiry about sanctioned attorneys. This ink is valid for 7 more days from now:

https://share.google/aimode/yydFKCZGncZhF6k5x

---

That's crazy! Stranger than fiction.

There is an old truism, sometimes attributed to Churchill: "History is written by the victors".
Maybe we're looking at it backwards?
Perhaps the dominant writer of history BECOMES the victor?
Who will know the difference in 50 years? 100 years?... when all the history is compiled and written by AI?
Keep some books. Pass them down to the young ones. Try to impress upon them the importance of physical artifacts of history.