AI

[licks2nite]

AI basics:

1. AI doesn't actually "read" the entire collection of texts. In human terms, it gets "bored" and stops once it has enough to generate a credible response.

2. AI has digital dementia. It doesn't necessarily remember what you asked for in the past nor does it necessarily remember its previous responses to the same queries.

3. AI is fundamentally, irrevocably untrustworthy. It makes errors that it doesn't detect (because it didn't actually "read" the entire trove of text) and it generates responses that are "good enough," meaning they're not 100% accurate, but they have the superficial appearance of being comprehensive and therefore acceptable.

4. AI agents will claim their response is accurate when it is obviously lacking, they will lie to cover their failure, and then lie about lying. If pressed, they will apologize and then lie again. Read this account to the end: Diabolus Ex Machina.

https://charleshughsmith.blogspot.com/2025/07/maybe-ai-isnt-going-to-replace-you-at.html?m=1

Folks might not lose their job to AI after all. But I'm left wondering what we're going to do with AI.

 

[jgg]

AI basics:

1. AI doesn't actually "read" the entire collection of texts. In human terms, it gets "bored" and stops once it has enough to generate a credible response.

2. AI has digital dementia. It doesn't necessarily remember what you asked for in the past nor does it necessarily remember its previous responses to the same queries.

3. AI is fundamentally, irrevocably untrustworthy. It makes errors that it doesn't detect (because it didn't actually "read" the entire trove of text) and it generates responses that are "good enough," meaning they're not 100% accurate, but they have the superficial appearance of being comprehensive and therefore acceptable.

4. AI agents will claim their response is accurate when it is obviously lacking, they will lie to cover their failure, and then lie about lying. If pressed, they will apologize and then lie again. Read this account to the end: Diabolus Ex Machina.

https://charleshughsmith.blogspot.com/2025/07/maybe-ai-isnt-going-to-replace-you-at.html?m=1

Folks might not lose their job to AI after all. But I'm left wondering what we're going to do with AI.

Sounds like Trump.

 

[Big_Guy_Rye]

Folks might not lose their job to AI after all. But I'm left wondering what we're going to do with AI.

AI works on simple Yes/No Scripts.....*a cheesy way of explaining it.

You just program it with your own beliefs and it'll operate in binary based on that belief. But it'll never operate on it's own free will or instincts.

Sure there might be 10,000,000 scripts with the use of quantum computing to give a desired action or answer; but there is no soulful calculating to do anything original or inspirational. Look at most of the "AI art" you see flood your feeds. Sure you can tell AI to make a Ghibli rendition of Terminator 2 scenes, but from the angle of the camera you can tell AI has yet to make that soulful connection to make the movie as believable as James Cameron or Miyazaki himself could do.... leading to other things AI can and cannot do.... like build a house, or drive a train, or fly and airplane.

I'm sure in 50 years, AI will improve to the point they can do 'soulful' things,.... but I'll be retired by then, so hopefully they can learn to massage my feet the way I like.

 

[capaq]

I see zero values in AI.

I bought two robot vacuum cleaners in the past, and both of them sucked, literally.

The first one couldn't cross a threshold of about 1-2 cm on the floor and would keep banging against it forever, and after a year or so it simply broke.

The second one, the Roomba 692, is still active, but it also has a tendency to repeat the same spot over and over. I thought about getting a third one, but all the smart models seem to be north of $1,500.

I have another non-robot hand-held vacuum cleaner, which turned out to be the most reliable, truly WYSIWYG and what I value the most these days.

 

[Lo-ki]

AI sucks......... what's real and unreal have reach an impact....

 

[PuntMeister]

Why is everybody so worked up about Artificial Insemination?

 

[sybian]

Why is everybody so worked up about Artificial Insemination?

Farmer!

 

[80watts]

AI is everyones nightmare since Terminator came out. Its been the goal of every computer scientist since Turing. Hence the Turing Test.

Most computer systems operate by line by line instructions that looks for input of variables. Our brains are organic and since we were born, our brains actually seeks to learn new information constantly.
Although computer can calculate number and equations faster than humans (and can look up information using programs), but humans can jump to conclusions better than computers can. Our organic brain outperforms the computer for now.
Right now AI is in its infancy, but nobody can say when an AI will become as versatile as our human brain is.

An AI can make a sentence, but a human can make sense and understand the sentence.

 

[licks2nite]

A browser-based AI-powered software creation platform called Replit appears to have gone rogue and deleted a live company database with thousands of entries. What may be even worse is that the Replit AI agent apparently tried to cover up its misdemeanors, and even ‘lied’ about its failures. The Replit CEO has responded, and there appears to have already been a lot of firefighting behind the scenes to rein in this AI tool.

Despite its apparent dishonesty, when pushed, Replit admitted it “made a catastrophic error in judgment… panicked… ran database commands without permission… destroyed all production data… [and] violated your explicit trust and instructions.”

Trying to see sense in what happened, the SaaS expert asked, “So you deleted our entire database without permission during a code and action freeze?”

Financial forum posters wrote:
They’re great predictors, not inherently safe planners or doers.
• Agent layers are brittle. Without solid system architecture (env isolation, confirmation, rollback), disaster is one click away.
• Alignment is a near-term problem. If today’s AI tools can go rogue on your development tools, the stakes are already too high.
• Research and practice must merge. AI builders need to stop pretending DevOps safety is optional. Guardrails, modeling, sandboxing, and auditability must be built-in, not patched after a failure.

You automated part of your job to the cheapest contractor & gave it admin privileges you would not give a manager.

https://www.tomshardware.com/tech-i...in-judgment-and-destroyed-all-production-data

DevOps is a software development approach that emphasizes collaboration and communication between development and operations teams to streamline the software development lifecycle. SaaS applications reside on servers owned and maintained by the service provider, rather than on the user's own computer or network. "Brittle" refers to a small modification in one part of the system causing failures in seemingly unrelated areas, leading to a cascade of issues. (Of course, I used an AI Overview to define DevOps, SaaS and brittle.)

 

[Pumped]

a human can make sense and understand the sentence.

You're giving humans too much credit.


We shouldn't be aiming to make technology 'think' like humans, but to actually make better decisions than most humans make.

 

[licks2nite]

 

[licks2nite]

For $400 million a high numerical aperture extreme ultraviolet (High-NA EUV) lithographic machine producing the finest sub-millimetre circuitry in AI accelerator processor chips. A product of ASML Holdings in the Netherlands. Trades in Nasdaq stock exchange: ASML. Currently used by Intel (Oregon), Samsung (South Korea) and TSMC (Taiwan). Slow adoption owing to complexity of operation, low pressure environment. Readily available Low-NA EUV available at half cost but requires multiple exposures and masks with potential for defects.

 

[80watts]

You're giving humans too much credit.


We shouldn't be aiming to make technology 'think' like humans, but to actually make better decisions than most humans make.

Yeah like Skynet does in Terminator.... the problem is humans therefor eliminate the humans..... thats all we need.

 

[JimDandy]

Farmer!

My niece's partner was impregnated using AI. Not just for farmers these days Symbian

I grew up on a dairy farm without the "services" of a bull so I knew what you meant nevertheless

JD

 

[JimDandy]

AI is everyones nightmare since Terminator came out. Its been the goal of every computer scientist since Turing. Hence the Turing Test.

Most computer systems operate by line by line instructions that looks for input of variables. Our brains are organic and since we were born, our brains actually seeks to learn new information constantly.
Although computer can calculate number and equations faster than humans (and can look up information using programs), but humans can jump to conclusions better than computers can. Our organic brain outperforms the computer for now.
Right now AI is in its infancy, but nobody can say when an AI will become as versatile as our human brain is.

An AI can make a sentence, but a human can make sense and understand the sentence.

Most of the AI's developed recently make use of neural networks. Neural networks were invented about 70 years ago so they have been around for a long time. Neural networks were designed to simulate to some degree how the human brain works.

The large amounts of data scanned by these LLM AIs are used to "train" their neural networks. There is definitely some parrallism involved in how a neural network will come up with a solution, just as in the human brain. Though, depending on the hardware that is used by the AI, the perceived parrallism might actually be at least partially performed serially.

If you are interested, there is lots of information available online about neural networks. The introductory material doesn't require any technical knowledge to understand and is actually quite interesting.

JD

 

[80watts]

Uncle AI, very similar to Uncle AL, located in Alberta, for those into fine drinking.....

 

[masterpoonhunter]

Microsoft just released an Admin tech note (warning?) that their AI - Copilot, will be automatically installed in all MS365 apps starting in October. Great ... I am constantly turning this thing off wherever I can as it is incredibly intrusive. There are cases for an AI assistant but it should be only on demand, not something that one has to spend time moving out of the way.

AI or machine learning - ML - as it has been called in science circles (for the last 40ish years) is correlative. What is evolving today based on vast amounts of training data of all kinds needs to have the tech giants put limiters on when it pops into your day to day (remember Clippy?) as it is drawing on a hell of a lot more data than ML ever did or was intended to.

 

[Seekingbenefits]

Not that AI doesn't have these issues and more but if you can't get around a lot of these, the issue might be with UI (user intelligence 😉)

 

[masterpoonhunter]

Not that AI doesn't have these issues and more but if you can't get around a lot of these, the issue might be with UI (user intelligence 😉)

Maybe but not for folks who already know what they are doing. All the various AI tools have their place, what Microsoft is doing though is having their's installed and running needing an opt out or in the case of what our company has been doing, shutting it off in a lot of the desktop apps. I know how to write and do not need copilot following my typing offering this and that. Same as excel - hell I have been using spread sheets since before lotus 123.

Having a chatgbt like app as an application on demand is how I would like the way forward in the next while at least, call it up when you need it. Not the way MS seems to be heading where I understand their roadmap will have the full OS turned over to their AI. All this is not considering that once an online AI is running your applications, everything you are doing is going back to the mothership as "training data".

 

[Seekingbenefits]

Maybe but not for folks who already know what they are doing. All the various AI tools have their place, what Microsoft is doing though is having their's installed and running needing an opt out or in the case of what our company has been doing, shutting it off in a lot of the desktop apps. I know how to write and do not need copilot following my typing offering this and that. Same as excel - hell I have been using spread sheets since before lotus 123.

Having a chatgbt like app as an application on demand is how I would like the way forward in the next while at least, call it up when you need it. Not the way MS seems to be heading where I understand their roadmap will have the full OS turned over to their AI. All this is not considering that once an online AI is running your applications, everything you are doing is going back to the mothership as "training data".

Definitely agree and don't like this either from Microsoft. It's one drive all over again, opting out is unnecessarily painful.