I bet AI can't come to my home and fix the A/C or replace a toilet!

I bet many of us have fixed stuff ourselves with advice we got on the internet including AI. It is easy to take a few minutes and confirm such advice through a non AI source. And given that it is known that AI has errors, everyone should take time to check.

The more important the info, the more non AI oversight one should provide. If you are writing a personal letter or moving furniture around, that's one thing (double check room and furniture dimensions yourself). If you are writing a legal brief or a medical plan of care that is quite another.

I see in this thread that many people are (justifiably) uneasy with chatgpt, even as others point out the positives. It is fine to point out either side, but this is not a contest, both points of view, and each of our opinions, have a valid place here. You can read all of what is written and come to your own conclusions, you do not have to have your conclusions provided to you.

I'm unhappy with the misnomer, "chat" if it's being used in concrete ways.

"Hi, I'm you're new best friend. Let's see how I can use you."

Lucille, we all know that opinions are only as good as the information one uses to form them. In that way, AI bots give you what is basically an opinion based on information that the bot has no way of knowing is true or not. Their proprietary algorithms determine what sources they pull from, how many sources they compare and which criteria are used to shape their answer. Do they have access to raw data or are they solely dependent on what has already been processed? They have no way to evaluate facts within any experiential context, nor can they watch a speech or a press conference and glean insights from the speaker’s demeanor, hesitations, or other aspects of a human mien. As Chispa said, “garbage in, garbage out.” I would say that it’s more a case of ”unknown in, unknown out.” And in too many documented cases the ”information” an AI bot offers is simply fabricated.

My biggest fear is that people will stop even trying to solve problems, evaluate products, or construct arguments for or against an issue. When the digital universe of entertainment became ubiquitous some feared reading would become a quaint habit as people could really get along without it. That has come to pass if you look at trends in reading scores and habits. I’m worried that, without daily practice, people will lose the art of abstract thinking. That’s enough to keep me awake at night. 🤷‍♀️

AI is a powerful tool if people use it properly and understand its limitations. A good article to read: Researchers Explore Mutual Benefits of AI and Science - MIT School of Science

My two college professor friends who teach logic and philosophy to intro level undergrads have finally after two years, succumbed to the fact that their students are using AI for every assignment. They are both teaching how to construct and understand what an argument is so the students can read and evaluate what AI is putting out. They have stopped all papers and any kind of writing assignment. They only give in class tests. And they tell their students exactly what questions will be on the test so they no longer have to deal with students and their parents saying there is any kind of unfairness in what is expected of the students in way of preparation.

We will all get dumber. More and more jobs will be gone. Though, I don't doubt there are still some uses to it in the field of medicine. I'm reading a good article now in the New Yorker about AI and medicine.

It would be interesting to set up a test with a good layout design dilemma. Have three really good Houzz designers come up with layouts - Jan, Diana, Beverly. Then have two different AI programs suggest layouts and see what the results are. Why? I think it is important for any industry to know what they are up against with AI. The pros and cons would be instructive to us all.

" Elmer, my husband is an Ivy League trained physician and in his opinion—in its current state— AI tools used by the general population are garbage

Did I say anything about AI tools used by the general public? I didn't refer nor intend to involve such things. The value of custom expert systems in the hands of experts has been long established. It's not novelty or fluff.

There's a series in the NY Times called Well. I think the Washington Post has something similar. Articles that describe the difficult paths some individuals had to follow until the elusive but ultimately correct diagnosis for their undefinable condition was found and the right treatment applied. Sometimes the breakthroughs come from residents and fellows who happened to remember something very obscure that could apply but hadn't been considered. Ask your husband to reflect on how some diagnoses can be impossible to track down.

That's what medical (in this case) systems are intended for. It's not Medical Google for the general public (no access would be provided), nor a Google for Docs or shortcut to be used in every patient instance.

food's comment above is exactly on point. He understood what I was talking about, others didn't (including one predictable one who either has trouble following along or likes to intentionally redirect to misconstrue what I've said). I won't waste time on replying to nonsensical comments.

I didn't mention ChatGPT in my serious comment, I didn't mention other things that are presently experimental and offered to the public as novelties. I know nothing about it, but I don't go to bed wearing an aluminum foil skull cap to protect me from the future and the bad guys in black. Read up about the Luddites in 19th century England for a frame of reference. "The Future" has never been something to fear.

Yes I like it a lot especially when looking for facts from opinions

I haven't read all the replies yet, but I had to respond to this. Do not trust AI to give you facts. It's very often wrong.

AI has uses. I do not use it much. I started using it at work to help me write specifications and release notes. It was helpful to reword my stuff to sound better. Then I retired and now do not have any use for it. It cannot be trusted if you're researching something.

Has anyone else noticed that websites and apps on our phones do not work well anymore? I am convinced that these companies are using AI and it is making things just not work.

I’ve noticed the same thing Jsk.

@palisades_ What you said doesn't make much sense to me - "Yes I like it a lot especially when looking for facts from opinions". Do you mean that you want to separate facts from opinions? I can't figure out how you would get facts from opinions.

“I haven't read all the replies yet, but I had to respond to this. Do not trust AI to give you facts. It's very often wrong.”

You wouldn’t make this hasty comment if you had read through the thread in my other posts where I strongly caution to validate the AI sources, and my experiences with it has been positive. But thanks for the reminder for the rest of the crowd who have had bad experiences with AI, which is understandable at this early stage of the technology, and until more regulations are in place and AI technology continues to advance then more minds will be changed. Nevertheless, people need to be vigilant with any information presented to them AI or not. If you don’t feel comfortable to use AI as a tool at this stage, then don’t, just wait for a better time in the future.

This is where the future is going - Recommended read for everyone: Researchers Explore Mutual Benefits of AI and Science - MIT School of Science

" I strongly caution to validate the AI sources "

If you're going to spend the time to double check everything that AI tells you, why not just gather the information yourself? How is AI saving you any time or mental work?

Tererosa & TV, I participate in a professional forum which sometimes I like to use AI to verify certain information presented in a discussion. Then I can cross check with references online from AI source or my own references on shelves. It’s quite fast and efficient when AI gives links to the online sources that I have privileged access.

" If you're going to spend the time to double check everything "

I suspect you've never been a team leader directing and supervising the work of more than one or two helpers. An equivalent comment for managing a large team situation, equally silly as this comment, would be:

If you're going to spend the time to double check and review everything your team does, why not just do all the work yourself?

Whatever, dude. Your response makes no sense, but you do you.

“If you're going to spend the time to double check everything that AI tells you, why not just gather the information yourself? How is AI saving you any time or mental work?”

TV - That question leaves me wondering how much you’ve tried to use AI. Where google returns a list of sources that may or may not contain your answer, AI goes in and sees if the actual answer is there. And provides related discussion, if the exact answer is not found. And discusses differing answers and reasons for the discrepancies, and shortcomings of the responses. Yes you still should verify and evaluate the sources, but it can give you a much better starting point.

I do think AI can be useful for docs when used in tandem with their own sleuthing skills in trying to find the zebras. It can be used as another "brain" in the room, another mind that can dig through and piece together odd collections of symptoms that might not occur in diagnosing outlier patients. Docs, I hope, can sit through what AI generates and figure out whether the AI suggestions are useful.

" AI goes in and sees if the actual answer is there. "

That's what it's supposed to do, but it doesn't do it maybe 25% of the time, if not more often - and Palisaides said they caution to "validate the AI sources". Because of that high hallucination rate, everything AI tells you would need to be double checked if you want to make sure AI is telling you the truth.

I wouldn't trust a system that's wrong at least 25% of the time without double checking everything; relying on it to tell me what information isn't available (rather than just making it up) isn't something I would do or recommend anyone else do. I think people are being gaslighted about the ridiculously high error rate.

“AI goes in and sees if the actual answer is there. And provides related discussion, if the exact answer is not found.”

And demonstrably bolsters the ”discussion” with manufactired ”facts.” Not all the time but enough of the time. Knowing a batch of frozen dinners from Company XYZ had a life threatening bacteria in 25% of the units, would you

1. take a chance and eat one purchased at random?

2. take the one you bought to a lab to be tested before you ate it? Or

3. Say to yourself, Gee, I think I’ll avoid that brand and buy something else? 🤔

Sorry but I find that analogy really weak. We’re not talking about life and death we’re talking about an efficient way to narrow down a starting point. The importance of what you’re trying to find combined with the nature of the question will drive how far you should verify. To me saying it’s not a useful tool is like saying google isn’t useful because it may not put your preferred answer at the top of the page.

Google is not as useful as it could be not because they don't put my preferred answer at the top of the page, but because they don't put the BEST answer at the top of the age. They put the entity that pays them the most at the top of the page, skewing the results for profit. This is one of the cancers of late-stage capitalism.

“one of the cancers of late-stage capitalism.”

What else do people expect when Google is free for anyone to use?

“but because they don't put the BEST answer at the top of the age. They put the entity that pays them the most at the top of the page, skewing the results for profit.”

From google: “No, Google does not place the entity that pays the most at the top of the entire search results page. Separate, labeled advertisements appear at the very top and sometimes at the bottom of the page. The core "organic" search results, which appear below any ads, are determined by Google's complex ranking algorithm and not by payment.”

You need to utilize google advanced search features to narrow down your searches. You need to put a little more work in to direct the tool (google) to work more efficiently for you. Otherwise GIGO sort of as mentioned earlier.

This was in the NYTimes and mentions one of the dangers of Chatbot is that it tries to be your 'friend.' The headline reads "Next Time you consult an AI Chatbot, remember one thing" - they are not neutral.

https://www.nytimes.com/2025/09/26/well/is-ai-validation-healthy.html?campaign_id=158&emc=edit_ot_20251013&instance_id=164185&nl=on-tech&regi_id=164274100&segment_id=207837&user_id=f822efcb26f7629b85f20e7547b93d2d

^ thanks for that article. It really explains a lot about what is happening these days. I will keep my answer seeking (for the most part) contained to my human friends. I'm not looking only for answers that agree with me.

I use it for all sorts of things. I consider the results to be like a frist draft. A lot of what it does, especially with anything creative, feels very synthetic and cold to me.

I'm sorry to share this observation - I think many of you don't understand what artificial intelligence is or isn't. I think many comments suffer from misunderstandings.

Artificial intelligence involves so much more and goes very, very far beyond typing a question into ChatGPT and noodling over the response. Would you flip through a Red Cross first aid book and decide you know enough to do open heart surgery?

Here's an early but a technically trivial but incredibly important application of machine learning everyone knows - GPS navigation. First on separate devices and now in cars and from smartphones. Do you naysayers find that evil and nefarious?

For those more interested in learning about it than carping with a lack of understanding, there are online "Elementary Introduction to AI" classes. I took one some years ago. It was fascinating and most useful. I suspect the yellow book series "For Dummies" has an AI title.

Do you wanna know or just wanna carp? The word "ignorant" isn't really a pejorative. It simply means "not knowing". Ignorance isn't just bliss. It's flat out ignorance

The OP's question was specifically about ChatGPT.

FWIW, I mentioned this upthread, and think it bears repeating - you can literally ask ChatGPT. 'what was incorrect about that answer?', and it will tell you.

I think it's unfortunate that 'hallucination' is the term of art now, when it really just means 'wrong'.

Here's a demonstration, starts @ 13:27:

My initial thought about Chatbots is that it's a "solution" only for those with limited skills.

I have two kids with advanced degrees in machine learning and understand very well how the commercial versions of AI work. And another who writes content for a google subsidiary. The public statement about search results is not the complete answer.

" The OP's question was specifically about ChatGPT. "

And, so what do you have in mind by saying this? The discussion has ranged beyond the starting point, as most discussions do.

Chatgpt is FREE, readily available and seems to be the most complete, has the broadest knowledge base, for an AI discussion. The topic is about AI and where we stand using it.

I use ChatGPT primarily to refine how I communicate — to ensure my message flows well, avoids offense, and sometimes gently shifts focus or uses a touch of psychology for persuasion.

" I have two kids .......understand very well "

Yes, but do you? Your earlier quote of your husband's comment didn't suggest so.

I'm sure you're very proud of your kids and have good reasons for it. Raising kids who worked hard to achieve in-demand employable skills and then found jobs to use them is not something that happens on autopilot.

Fake rock climbing dog by new Sora video application: A.I. Video Generators Are Now So Good You Can No Longer Trust Your Eyes https://www.nytimes.com/2025/10/09/technology/personaltech/sora-ai-video-impact.html?unlocked_article_code=1.tk8.pFtQ.tFU5DCzQsnar&smid=url-share

I've been watching FAKE Yeti stories....they're so kind and caring....working with humans like they do....🤣🤣🤣

Note: It's a current trend among Gen X'ers to have a Yeti sticker on their car saying "they believe".

I have talked about this before, but there are Facebook stories which appear in my feed, which are entirely fictional. But presented as if they actually happened. Because I follow tennis, the modeling shows me content with tennis players. It all now has a pol i t i cal slant. So and so ( allegedly) either being friendly with, or arguing with, well known figures.

It is beyond bizarre. All produced with AI. Looks real too. I have watched some of it, so I am being shown this garbage frequently now.

How many people watch this garbage and believe it? That is what truly scares me.

Elmer, interesting that you offered GPS as an example. Nearly every time I use GPS and I use it often if I have never been somewhere before, it takes me on a route that I would not use to get out of my neighborhood, for example. And when I look at the exact route it takes, I can find much better ways to go though I do use it for the final destination which isn't always correct.

I think AI can be useful when taken at it's face value and then decided upon it's usefulness. Maybe that is how all AI should be looked at - with skepticism. If you use ChatGPT (I don't) and then have to ask what is wrong with the answer, I think that should be a red flag.

Just my opinion.

Ally De, I, too, see fake AI content all the time. And now most comments are people pointing out what is fake about it. There are a whole slew of videos showing a baby (always the same baby) sitting outside and then being attacked by a cat or a coyote or a dog or a person usually being saved by a dog. But these are ludicrous and who cares? Are people that hard of for actual entertainment that this is how they are spending their time? I really can't understand it. Life is too short to invest in this kind of nonsense.

AI is a double-edged sword.

Murraysmom - isn't it wild? Yes I agree, I am now usually seeing comments from people pointing out that the story or photo/video is fake. That does give me small consolation.

I also get fake/made up stories about the British monarchy. I made the mistake of reading a link about the whole Harry/Megan thing, and I was then over-run with fake AI stories/photos about the entire monarchy.

If you make the mistake of clicking a link, you're almost guaranteed more of the same.

But I suspect they even track how long you scroll/linger over something in the feed even if you don't click a link, because I've noticed if I stop for a few seconds to read something in my feed I see more of the same even if I never clicked on it.

Facebook is garbage now. I presume anything they inject into my feed is false.

I'd like to give it up entirely, but my HOA uses Facebook a lot....and I have a few friends from back in the dark ages who I do enjoy seeing their pictures and keeping up with them that way. I've figured out how to get to the "just your friends" feed, which does seem to slow down the fake "news"....somewhat.

" I can find much better ways to go "

Well, maybe, maybe not.

First keep in mind that there are settings that you may or may not have used. Do you want the shortest distance or the shortest time? On highways or not? Etc.

Also, it uses various criteria like those settings, current or stale traffic status it may have, speed limits for each road that can be taken (and assuming that they're accurate and you will be able to drive at the posted speed on each road), etc. to iteratively try different routes to ultimately come up with what it presents as the best alternative.

All it can do is adhere to the algorithm and that can lead to odd results. If one way is 50 feet shorter than 2nd place but you know that the route chosen can have a lot of trucks, or momentary delays here or there, then your knowledgeable choice may be better than what's presented. I was once in a French wine region and my GPS routed me to take a dirt road through a large vineyard, in order to get me to a restaurant in the neighboring town. I stopped where the paving ended and turned around. It was a good laugh and I was able to redirect to more suitable choices of roads

They're not omniscient nor infallible. And don't need to be. They present useful information that can be helpful. I just used 3 different navigation systems on my phone to compare guidance for a trip I take regularly. Apple Maps, Waze, and Google maps (which with the built in system in my car is my co-preference). The answers were different and with each set to choose the shortest time for the trip and varied by 40 minutes from shortest to longest. All would get me to the destination. The difference likely was because of variations in the traffic congestion data each was using. That didn't matter to me significantly because I know such data can be stale or easy to overstate and understate. What would be useful if I were making the drive now would be current indications of slow downs, accidents, construction, etc. so that I could add my own judgement to the routing or tell the system to take an alternate route. And several of them alert users to alternatives on a real-time basis.

It's a good example to consider because even sophisticated systems are likely to not be perfect, not infallible, but very helpful in pointing users in the right direction for whatever the task is.

I was sitting in the steam room today, guy starts telling me about "Constellation AI". Says I should check it out. I Googled it:

AI Constellation is a large-scale AI collaboration technology that enables AI systems to engage in mutual discussion and correction, thereby generating solutions from diverse perspectives for problems that are difficult even to define.

If I find the time...I'll look at it a little closer...

A coed steam room? That's something new.

I googled it too and it appears to be a project of a small IT company. Perhaps it's something noteworthy, or, perhaps and maybe more likely (in my opinion based on other opportunistic things I've seen) a small fish trying to make noise and attract attention.

I heard a good use of AI today: meal planning.mYou give it the things you like to eat, I guess give it your dietary restrictions too, then ask it to generate a recipe and a menu around it, and a shopping list, times 7 or however many days you want to plan for. You take the shopping list to the store, then you have everything you need to make your meals for the week. Of course if salt is on the list, and you already have salt, you won’t buy it. I suppose you can edit the list.

I used to subscribe to such a service, no AI involved, and I really liked it. Some things I never would have thought to tell AI that I wanted to eat, but this service sent them to me, and 95% of the time I liked it. My kids were eating kale which floored me.

@Elmer J Fudd... A coed steam room? That's something new.

No nudity. Vasa fitness centers. Really nice place, we have 5, they're a chain. 215 + photos

" I heard a good use of AI today "

There's a difference between AI and generative AI. There are a lot of good uses for AI, and even things photo recognition to help with a search use AI - like on your iPhone if you search your library for a picture of dogs. But generative AI is a different thing altogether, where it creates something out of input that might be wrong - either factually wrong like with legal work, or visually wrong like a person with 6 fingers or four eyes.

@Toronto Veterinarian, I agree. In fact I was skimming through some of the comments yesterday and wanted to comment about how some things can be trusted (to a degree) and some not much at all. Your legal work example is just one that has surfaced recently and is a big deal in that world. I have found several incorrect identifications and summaries in the AI blurbs based on my own knowledge but i just scroll down to the next items. I never just take the AI summary as the final answer.

One of the state journalists I enjoy reading and highly respect (he’s been a Pulitzer recipient for a couple of his in-depth works) had a very interesting article about his recent experiences yesterday. He googled his name. It was really annoying to read what he was purported to say and write vs what he actually did write. I’m sure he is highly annoyed. So just imagine if untrue statements were attributed to you and you were now looking for a job for example. Not good. I can try to link his article here if anyone is interested or paste most of it in a summary.

For fun I googled my name and while I didn’t merit the usual AI summary at the top, it was interesting to read and see all the errors, which i will never bother to correct in my lifetime. Mostly phone numbers (it listed around 20-25 and I may have had 5-6 true personal numbers not counting work office and cell numbers) but several addresses I had no idea about.

I’ve been in the fence about ChapGpt. Nicole I can see where you find a good use for it and i would probanly do the same if needed. I’ve for sure used an image search many times on both the Google platform and on several apps I routinely use for plant, insect, and bird identification (Picture This, Merlin, Picture Bird/Insect/Nature)

Interestingly, Jon Stewart had a segment last Monday about the rise of AI and what's driving it, along with its impact and negative aspects some people might not have considered.

carolb -- THANK YOU! A quick look at why AI is scary.

I'll be looking into the Center for Humane Technology. Sounds hopeful.

I recently had a very positive experience using ChatGPT.

I had been sitting around airports and on long haul flights for days. I came back to my room after some business meetings to change for dinner. I was lying down when I felt a sudden sharp pain in my lower leg. I thought it was a cramp so I stood up and walked on it, and drank some water. It was not getting better and it was very painful. I have had three kids with a total of one Tylenol so I think I have decent paint tolerance; this was intolerable.

Worse yet, I have a history of DVT (deep vein thrombosis), and given how much sitting I had done the last few days, and the fact that it was in a lower extremity and only one side, I found very concerning. A clot can travel to the heart or the brain and make quick work of you. I considered calling 911.

I asked ChatGPT. I entered my symptoms and quick history, etc. It said it was unlikley a DVT. I then asked, how can i rule out DVT? It asked me to be more specific about where the pain was and describe it. It was on the front of my leg, mid-shin. That fact enabled ChatGPT to rule out DVT, along with the lack of swelling or discoloration. Moreover, it asked a few more questions and suggested it may have been caused by walking in heels more than I usually do (true, prob about 2 miles).

But for ChatGPT I would have called 911 and that would have been a whole rigamarole. My leg bothered me a bit later that night in bed but was fine after that.

ETA - the other options i had to research this question were too verbose and did not allow me to customize my request by adding my history, etc. ChatGPT was the quickest way to find an answer, by far.