Open Source AI is Amazing

[melk]

The one thing that sucks about this being "a single .jpg" is that, being an image, none of your information can be selected as text and copied/pasted. The only thing that sucks more is trying to craft a half-decent email signature!

MacOS / iOS can select text from images. I would have thought Windows/Android can also?

 

[gman2304]

I’m a pioneer in Artificial Intelligence. I’ve been using it my entire life!

 

[Brons2]

I don't agree that the open source AI is good. The commercial AI models are much better. I've tried and gone back to the commercial ones, i.e., Claude, ChatGPT, Copilot, Gemini.

I suppose that you could train your own open source model that was just as good as the commercial sites if you had the right hardware to support it. The tools are out there for anyone to access and set up your own.

It's fun to play with the local models if you are a PC enthusiast like I am, but, for serious work I'm going back to Claude or ChatGPT. If I have to upload work data, then Copilot because we pay for it and the data is not used to train the model per our license.

 

[el Nuke]

I'm not going to say that AI can go through your attic. That is actually proving a point that many are saying about AI, that it isn't going to replace everybody's job.

But it surely has many capabilities. Take this article from WaPo, talking about how the US military used AI in the strikes on Iran (not sure if this is behind a paywall or not, I believe this article can be gifted for the next 14 days):


The "system" referenced in the quote above was the US military's Maven Smart System, which was according to the article, built by Palantir & powered by Claude, whatever that is supposed to mean.

It appears to be behind a paywall, at least for me.

But the one paragraph I was able to read noted how the use of AI helped with the abundance of targets in the early hours of the conflict.

I wish I could read more. I would very much like to know whose colossal mistake it was to bomb the school. Was that the fault of the AI or a human?

Because if it’s a human, that human needs to be called out on the carpet. If it was AI, well, I guess nothing can really be done in that case…

 

[paulri]

You might need to set up an account, but the article can be accessed free. But no paywall.

It appears to be behind a paywall, at least for me.

But the one paragraph I was able to read noted how the use of AI helped with the abundance of targets in the early hours of the conflict.

I wish I could read more. I would very much like to know whose colossal mistake it was to bomb the school. Was that the fault of the AI or a human?

Because if it’s a human, that human needs to be called out on the carpet. If it was AI, well, I guess nothing can really be done in that case…

 

[OVERKILL]

What I would see as a concern of AI is a continuous dumbing down of people’s god given gift of researching, and critical thought and thinking.

AI surely makes me lazy, and gives information I likely take as fact, and the information may be absolutely false.


Kind of like video games of sports replacing actual sports. One misses out on the leadership, followership, pressure, and preparation one receives when playing actual sports, and receives none of those benefits playing video game sports.

This. I regularly see AI-generated content presented as absolute fact by people that know little to nothing about a subject. It's lazy and rude, but at least makes it easy to identify who to avoid interacting with further. I saw a Canadian politician yesterday quote AI staffing figures for Canadian nuke plants that were so far from accurate it was laughable.

What would have typically required at least some basic level of research is now being replaced by a prompt and a convincingly worded response that's just accepted as "the truth" without even a glimpse at the source(s). Manifesting alternative realities based on AI fever dreams and slop.

That said, for hobby-level programming, it's great for efficiency. Saves huge amounts of time.

 

[IndyIan]

This. I regularly see AI-generated content presented as absolute fact by people that know little to nothing about a subject. It's lazy and rude, but at least makes it easy to identify who to avoid interacting with further. I saw a Canadian politician yesterday quote AI staffing figures for Canadian nuke plants that were so far from accurate it was laughable.

What would have typically required at least some basic level of research is now being replaced by a prompt and a convincingly worded response that's just accepted as "the truth" without even a glimpse at the source(s). Manifesting alternative realities based on AI fever dreams and slop.

That said, for hobby-level programming, it's great for efficiency. Saves huge amounts of time.

Yep, I do find many of the free options tend to give you the simplest answer it thinks you are looking for. Or sometimes spits out non-sensical statements on topics I know the subject matter, so it makes me a bit skeptical when researching something I know very little about. A follow up, "Are you sure?" sometimes give you a totally different answer!

 

[OVERKILL]

Yep, I do find many of the free options tend to give you the simplest answer it thinks you are looking for. Or sometimes spits out non-sensical statements on topics I know the subject matter, so it makes me a bit skeptical when researching something I know very little about. A follow up, "Are you sure?" sometimes give you a totally different answer!

Yeah, his said Canadian nuclear plants typically employ 300-500 people. This may be a reasonably accurate figure for a single unit US PWR in a competitive market, but ours employ about 10x that.

 

[Plumb Bob]

That said, for hobby-level programming, it's great for efficiency. Saves huge amounts of time.

For coding, hobby or otherwise, saves a lot of time. Of course, the watchful eye of a software engineer or programmer completes the equation.

 

[alarmguy]

I see it this way.
For decades now you can find "information" in search engines, its critical thinking to not rely on one source.
There are a lot of posts in here talking about AI but I don't know what AI they are using and I am very limited of the ones out there.

So to me, this talk about incorrect AI is lacking without knowing which one. No different then when just search engines were used, we would know which one.
For a consumer I have found ChatGPT excellent for what I needed and critically you can verify results with a search engine.

ChatGPT my wife and I find invaluable for simple questions without doing a search and REALLY invaluable (in my case) of dropping the entire copy and pasted results of my PSMA pet scan a year ago with prostate cancer, at Junipers (chatCPT) suggestion and offer to drop the results into the text box of the app while having a voice conversation with her. She ran down line by line exactly what everything meant and I can confirm it was what my doctor told me but forgot the details as we all may do after leaving the doctors office.

Regarding the exact technical aspects of the cancer results of what I learned, I did then do an internet search and she was spot on. The reason for this is I was trying to find out the reasoning for my rather "advanced" stage of cancer staging (and my family was wondering why too) even though only a little cancer was found. It all came together and was able to verify not only what two medical networks told me (one of them was DUKE) but a complete run down by Chat GPT and then piece by piece I found information backing up what all sources were telling me on medical sites.

Ok, agree, for less critical thinkers, I do "think" silly to rely on one source but wow, I was impressed and we have to remember, ALL this technology is still in its infancy.

 

[OVERKILL]

I see it this way.
For decades now you can find "information" in search engines, its critical thinking to not rely on one source.

Right, but you didn't tend to get a wholistic "this is the answer" from this traditional "search and piece together" method, which forced at least some level of critical thought and effort.

There are a lot of posts in here talking about AI but I don't know what AI they are using and I am very limited of the ones out there.

All of them. I've seen it with all of them.

So to me, this talk about incorrect AI is lacking without knowing which one. No different then when just search engines were used, we would know which one.
For a consumer I have found ChatGPT excellent for what I needed and critically you can verify results with a search engine.

I've proof-read AI prompted nuclear papers, done with ChatGPT, that had wholly incorrect information in them. The criticism applies to every model. The problem is a bit of a paradox, since in order to understand why the answer is wrong, you need to know the subject matter, and if you deeply know the subject matter, you are unlikely to be asking AI for the answer.

ChatGPT my wife and I find invaluable for simple questions without doing a search

And that's what people are doing. And many assume that "because it's from AI", that it's by default, correct. Like the aforementioned politician quoting employment numbers for Canadian nuclear plants. If you are going to the trouble of reviewing primary sources, you've at least semi-defeated the purpose of prompting the LLM in the first place.

and REALLY invaluable (in my case) of dropping the entire copy and pasted results of my PSMA pet scan a year ago with prostate cancer, at Junipers (chatCPT) suggestion and offer to drop the results into the text box of the app while having a voice conversation with her. She ran down line by line exactly what everything meant and I can confirm it was what my doctor told me but forgot the details as we all may do after leaving the doctors office.

Regarding the exact technical aspects of the cancer results of what I learned, I did then do an internet search and she was spot on. The reason for this is I was trying to find out the reasoning for my rather "advanced" stage of cancer staging (and my family was wondering why too) even though only a little cancer was found. It all came together and was able to verify not only what two medical networks told me (one of them was DUKE) but a complete run down by Chat GPT and then piece by piece I found information backing up what all sources were telling me on medical sites.

Ok, agree, for less critical thinkers, I do "think" silly to rely on one source but wow, I was impressed and we have to remember, ALL this technology is still in its infancy.

My best friend is deep into the asking ChatGPT health questions lately, in large part because it has a better bedside than his GP. He doesn't assume everything it says is absolute fact, simply looks at it as getting another opinion, which he can then ask his GP about at their next appointment. But people are inherently lazy, want to be told what they want to hear, and want what they want NOW. AI checks all those boxes. I don't think that's inherently a good thing for the bulk of the population.

 

[alarmguy]

...

My best friend is deep into the asking ChatGPT health questions lately, in large part because it has a better bedside than his GP. He doesn't assume everything it says is absolute fact, simply looks at it as getting another opinion, which he can then ask his GP about at their next appointment. But people are inherently lazy, want to be told what they want to hear, and want what they want NOW. AI checks all those boxes. I don't think that's inherently a good thing for the bulk of the population.

This is me, like your friend, my point is it's (right now anyway) like a search engine of search engines. I dont know how anything combing the interest is an absolute matter of fact. Used properly it's an invaluable aid and we are only at the cusp of what it can and will be.
Being informed is the best possible healthy thing you can do for yourself to ask proper questions both from your doctor and multiple sources of information that may confirm what CHAT GPT is saying. Then the bottom line is the physical doctor and possibly the reason for a second opinion.
Right now I am questioning a ANP who saw be the last week of June 2026 (not based on ChatGPT) but my own informed self. Wants to schedule a combination echo/stress test and gave the reasons. Family history tells me he should be doing more and he should have listened to me wanting a CT Angiogram. Though I didnt argue the point everything I read points in that direction though the echo/stress is also ok but my history CT Angiogram I think is better suited. I just finished messaging them because as thorough as I thought he was, his notes were not 100% correct.

First I am wondering why it took days to end up in "my chart"typically its the same day, 1st questionable thing, 2nd was the chest pressure I told him about he posted an incorrect note. Said I felt pressure Memorial Day weekend but not while playing pickleball and working in the yard. That was incorrect I felt pressure all weekend.
Also while typing you, I see a third mistake, we agreed and he sent a new prescription to increase my statin to 10MG!!!

Anyway, using the internet is what started me looking more closely at his notes and a CT Angiogram is certainly prudent test however the stress/echo is ok but CT Angiogram fits me better.

Maybe if time I will ask ChatGPT what "she" thinks. She talks awfully nice to me *LOL*
Anyway, think you know my personality as well. I dont take any one answer and when it comes to a critical component of the heart... well, I know a lot about them and testing do to family history and this recent visit is not adding up. Same type of smarts have to be used to judge ChatGPT answers.


Part of my Drs NP notes
"Today patient describes event 5/2026 with episodes of light chest pressure while walking his dog, occurring on 3 separate days but not with pickleball or yard work. He states symptoms resolved and have not returned since then. He has lost 34 pounds since last visit. He admits to being seen by PCP recently with recommendation for increase statin dose which she has not done yet, continues with Crestor 5 mg daily. He is followed by oncology for prostate carcinoma and continues with orgvyz. "

I just found a 3th error !! Thank you @OVERKILL for making me look more closely.
I am NO longer on the anti cancer drug Orgovyx (which he misspelled.)


Anyway, search engines and AI are aids but one needs to confirm with sources and then of course, in health issues consult with a doctor.