The way it's going is AI search, it tends to target the subject vs an all encompassing 40 pages of links that you get with Google. Clearly it's the future, so get used to it....

https://x.com/InnovationRapid/.../1926989060871975135

Question:
AI is taking over, listen, and understand. AI is out there, it can't be bargained with, it can't be reasoned with, it doesn't feel pity or remorse or fear, and it absolutely will not stop...EVER until everyone's job is dead... his job is dead, her job is dead, they're ALL DEAD! OK maybe that's a bit to dramatic....

Choices:
Yes I Grok it - or some other AI thing
Still google'n it, who cares what Sundar Knows
A quacking I DuckDuckGo!
Whats this AI crap
I'm wearing Tin Foil and never search the web

 

AI responses are sometimes useful. I trust but verify. Most of the time, I will still look for discrete links of interest. I don’t knowingly use google. I use brave.

It may well be the future, but the future is not now. Not fully ready for prime time.

It doesn’t matter. Googling turns up 1,000 AI generated bullshit webpages. AI search summarizes those 1,000 bullshit AI generated webpages. Either way, it’s all horse pucky.

I use Brave or DuckDuckGo instead of Google, but the result is essentially the same. Dozens of pages of increasingly irrelevant hits to sort through. The AI-generated summary at the top is usually easier to read and often even helpful, especially if my query was along the lines of "what is ABC?".

My biggest problem with AI is, I can't think about it without going immediately to Skynet. Or probably more realistically, "When is big brother going to start using AI to monitor us lowly subjects, or are they already doing it?"

quote:

Originally posted by 12131:
It may well be the future, but the future is not now. Not fully ready for prime time.

Agreed. Those who rely on so-called "AI" in its current state for searching/answering/researching really ought to read up on what a Large Language Model AI actually is.

Hint: LLMs aren't actually Artificial Intelligence... that's just a buzzword. They don't think or reason. They're eloquent because they're trained to mimic clear and natural-sounding sentences, but not intelligent.

Instead, they're just sequential word predictors, kinda like a fancier version of how when you start typing a text message your phone tries (but often fails) to predict/suggest which word or phrase you might want to come next.

It's not generating answers using fact/evidence-based critical thinking, weighing the validity and reliability of souces and drawing conclusions. Instead, they base their predictions off how frequently a word follows another word when pulling from big text repositories like Reddit and other sources.

Basically a word popularity contest. An automated groupthink text generator.

And importantly, LLMs are frequently wrong and have a habit of simply making things up, inventing bullshit answers which are stated confidently but which just aren't accurate. They're improving year over year, but still have a long way to go before they could be considered reliable.

I doubt whether LLM AIs specifically will ever be able to get to the point of being fully reliable, purely by their nature. I suspect full reliability won't come until we have created actual true Artificial Intelligence that's capable of thinking and reasoning.

Which is why I personally think it's a bad thing - and even dangerous - for everybody to be pushing LLMs and trying to integrate it into everything. Because it's not reliable and maybe never will be, yet there are tons of people out there (both young and old) who don't realize that and simply accept whatever the "AI" spits out as the truth because the computer says so and they've been conditioned to think that computers are "smart".


The generative image creation stuff is fun to play around with though.

This message has been edited. Last edited by: RogueJSK,

I asked a question the other day that I already knew the answer to, something close to this: "Is it true that MSM only reveals the race of a crime suspect when the person is White?"

The AI response was yes, that is how it happens most of the time, and then gave some supporting details. I was glad to see the truth for a change.

I'm not even to trust but verify yet.

AI can be nice, it can throw things together into an easy to read synopsis. The only problem is that it can easily be confidently wrong. I've seen it spit out information that is incorrect or even harmful. Your really have to fact check it thoroughly , and that takes longer than it would have taken to just google it in the first place.

For example, I've got a neglected lawn I'm getting back under control. It's primarily bermuda, but it has a whole lot of patchy St. Augustine mixed in. The bermuda is slowly taking back over, but I'd like to hasten it along so I searched for "kill St. Augustine but not bermuda". The AI overview confidently told me that to selectively eliminate St. Augustine while sparing bermuda I should used Recognition or Fusilade II. Further reading however shows that it got the associations backwards.

Those herbicides are actually used to control bermuda in St. Augustine or Zoysia. If I had blindly followed what the AI said I would have completely destroyed my yard, and killed the grass I wanted to keep while sparing the grass I wanted to eliminate.

Google is adapting. It also provides an AI answerwhen you google for a question or phrase.

It’s like the progression from pencils to fountain pens to ballpoint pens. And, yes, I asked AI what the progression of writing instruments were.

AI is great when you’re asking for specific information and it acts like a filter compared to Google ranking search results. I want it to aggregate and summarize the results for me. Like I asked the difference between marinating spare ribs in apple juice and pineapple juice.

I also challenge it from time to time by asking it. “Are you sure?” Or I would say I was expecting the answer to be such and such. It’s more efficient than clicking on links that come up on a google search.

As I’ve posted before: in the areas in which I have specialist expertise, Grok is 100% all the time.
So for technical questions, I am very happy to go right to Grok. Problem with my electronic keyboard? Grok gave me troubleshooting tips in order. Need a part for the dishwasher? Take a picture of the model number, upload it to Grok, get an answer right away and an estimate of repair time and if DIY-able.
I have tried no other AI programs. Not interested in supporting MS or Google or OpenAI.

^^^^ Doc, do me a favor and ask Grok the following question. Did the 1996 production Sig Sauer P229 come with a pinned front night sight?

GROK: "The Sour Pig came with a 229 watt night light that was a product of 1996 pines."

^^^^ Is that real xAI groking, or is it you monkeying around? Lol.

I am the infinite...

quote:

Originally posted by 12131:
^^^^ Doc, do me a favor and ask Grok the following question. Did the 1996 production Sig Sauer P229 come with a pinned front night sight?

Not the Doc, but here is the answer to your question:

quote:

The 1996 production Sig Sauer P229, introduced in 1992, typically came with standard contrast sights (non-illuminated) or, in some cases, factory-installed night sights, depending on the specific model and configuration ordered (e.g., law enforcement contracts). Night sights were not universally standard on all P229s during this period but were available as an option, particularly for professional use.
Regarding the front sight, early P229s, including 1996 models, often used a pinned front sight design for both standard and night sights. This involved a roll pin securing the front sight to the slide, a common practice in Sig Sauer’s Classic P-Series pistols at the time. If the 1996 P229 was equipped with factory night sights (such as SIGLITE or aftermarket tritium options like Trijicon), the front sight would typically be pinned, as this was the standard mounting method for P229s in that era. Screw-mounted front sights became more common in later models, particularly with suppressor-height or optic-ready configurations.
However, without specific serial number details or documentation (e.g., original box or manual), it’s hard to confirm definitively, as configurations varied. For instance, some P229s shipped with #8 front and #8 rear sights, while others had #6 front and #8 rear, depending on caliber (9mm, .40 S&W, or .357 SIG) and intended use. Night sights, if installed, were likely tritium-based, and the front sight would be pinned unless modified post-production.
If you’re verifying a specific firearm, check the front sight for a visible roll pin or consult Sig Sauer’s historical records or a gunsmith familiar with Classic P-Series pistols. Alternatively, I can search for more specific information if you provide additional details about the pistol (e.g., caliber or variant).

I still google it and try to find the original source material, or at least get as close as I can get. But that's getting harder and harder, as a lot of the content these days is actually AI generated, not just the search results. And quite a lot of it is incomplete or wrong.

I'm also not cool with the new trend of using AI to write stuff instead of doing it yourself. I recently attended a retirement party for a former boss, who had worked at the institution for over 40 years. Keep in mind, this was a college...an "academic institution of higher learning". A few people spoke, and they all admitted to using AI to compose their remarks...for a man who they'd worked with for decades. It just felt very cheap and impersonal to me...almost insulting.

I've heard rumblings at my current job of AI products available for composing Police reports, and have been shocked at how many cops are excited at the prospect. Do we really want a computer composing a document that is used to deny a person their freedom? Or on the flip-side, a document that you are going to be held liable for upon penalty of perjury? I want to know and have personally chosen every word I put in a document like that.

I think there are legitimate applications where AI can serve a valuable purpose, but overall our overdependence is just going to make us lazier and dumber, because that's what people do.

quote:

Originally posted by sigmonkey:


I am the infinite...

OK Skakespeare, back in the cage.

quote:

Originally posted by 92fstech:
I want to know and have personally chosen every word I put in a document like that.

110% agreed.

quote:

I think there are legitimate applications where AI can serve a valuable purpose, but overall our overdependence is just going to make us lazier and dumber, because that's what people do.

Also agreed.

A good example is something like generating computer code. LLM AIs are apparently quite good at generating programming code.

But that code still has to be tested to ensure it works. And it still has to be paired with a programmer who can troubleshoot it if the provided code doesn't work 100% correctly. (So it can cut down on the time needed to code something, and means we will require fewer coders overall, but it won't 100% replace coders altogether because they're still needed in the loop.)

That's an example of the correct way to use LLM AI. Take the AI's results, test/examine them, check them, and follow up with actual human thinking and reasoning if the answer is partially or fully wrong. Used to streamline/speed up the process, but not being trusted to do the entire process.

As opposed to asking a LLM a question, getting an answer that may or may not be accurate, and then just blindly/lazily accepting the answer with the assumption that it must be correct. Which is how many/most people seem to be approaching using a LLM AI, despite the fact that there's a significant chance that it's either inaccurate or even outright hallucinated/bullshitted altogether.

Like anything else, it's just another tool in the box, and it's up to the user to know which tool to pick for which job, and how to best use it.

That said, yes, for most things these days, I go straight to Grox and ChatGPT since they're generally a more efficient use of my time. If the results aren't what I'm looking for, I'll go search for myself.

quote:

Originally posted by 12131:
^^^^ Doc, do me a favor and ask Grok the following question. Did the 1996 production Sig Sauer P229 come with a pinned front night sight?

That's funny. Everyone knows no AI has yet to be programmed to be on the same level of expertise as you on Sigs.

"Can anyone tell me about the gun where this fuzzy close up picture of a nub is taken of?"

You in 0.23 seconds: "Yeah, that's a 1989 P228 Serial number series xxxx with an obviously modified whatchamacallit..... What else do you want to know about it?"