Postulation: Is anyone else concerned with the proliferation of AI?

[kapone32]

Sad. I hope you find something good soon!

I am going home and opening a Aracde/Sim Racing hub/Bar. My company has been good to me and buying Stocks for me for like 20 years and I was that guy that bought PC parts every time I got paid. Now it is just that I build the cheapest capable PC I can every Black Friday but I have everything I need. AM4 was great.

I worry about my Daughter, my Wife is caught up in the narrative so thinks what happens on Relationship realty TV is more important than what is going on if you are paying attention.

I had to add that this is the most one sided poll I have seen on TPU. I am glad that most people overwhelming agree with my position.

 

[ShrimpBrime]

I really don't understand you. You get stuck on a thought and refuse to see anything beyond that - even when you quote me, you refuse to see what I am saying.

First, those sensors do indeed look through the window - but to this point you just made, did I say that is how autonomous driving works? Nope!

I said, and you quoted me, "in the near future" then I went on to explain how things "will be" with AI working with IoT and how it "will be when" all cars are connected.

But apparently you refuse to accept what the future will bring because, according to you, that is "not even close" how autonomous driving works today.

I guess AI will never "evolve" beyond what it is today.

No. The sensors do not see through windows. It's sees the car as entire object. The radar on the side sees objects like a bat. No side cameras, but is pedestrian aware?? Nope, it is an object aware.

The autonomous driving systems do not see people as living beings in any way shape of form.

I study deeply these systems on Freightliner Cascadia class 8 tractors for a career and they update small revisions every 6 to 12 months.

Yep, that Cascadia behind you on the toll way has a minimum following distance of 1000 feet.

Near future..... maybe a decade from now maybe. These features on only on select make and models of vehicles. My very nice 2017 Passat does not have autonomous driving features, but does have brake assist.

 

[lexluthermiester]

Noticing the results of the poll. The vast majority seem to be worried about the potential problems.

 

[DaemonForce]

I'm not so much worried about AI itself but the problems it causes with stupid people. A lot of the problems we face in the tech world today are directly related to that weird and insanely awful 100-110 IQ range of bozos that can't separate fiction from reality. Meanwhile we have a lot of terminally online boomers that effectively function as bots in the aging and useless social media spaces. There are reasons we're seeing Zucc (and likely many others) getting ready to start his crashout arc. This is not a good moment for humanity.

 

[Bill_Bright]

No. The sensors do not see through windows.

Wow. Dude! You are just wrong! Some simple common sense must come into play here. Come on! The sensors that are located behind the rear view mirror or on the dash MUST look through the window! How else would they work? Even the automatic wiper feature on my Avalon used a sensor on the back of the rear view mirror (where it rests against the windshield) that looks through the windshield. The auto-highbeam detector on my truck looks through the window.

Are all sensors located behind rear view mirrors? NO! I never said or implied they were either. But many are and this is one reason you see windowshield replacement commercials talk about windshield recalibration after replacing a broken windshield.

Apparently it is not obvious but in the case of sensors, "looks through" and "sees through" does NOT automatically mean it does so with a camera.

The autonomous driving systems do not see people as living beings in any way shape of form.

Huh? What does that have to do with the price of rice in China in the summer time when it rains? Nothing.

Did I say anything about seeing people as "living beings?" NO!!! So why even bring that up? I don't get it.

they update small revisions every 6 to 12 months.

Whoopie do! Again, rice? China? Summertime?

The vast majority seem to be worried about the potential problems.

Right.

 

[ShrimpBrime]

Wow. Dude! You are just wrong! Some simple common sense must come into play here. Come on! The sensors that are located behind the rear view mirror or on the dash MUST look through the window! How else would they work? Even the automatic wiper feature on my Avalon used a sensor on the back of the rear view mirror (where it rests against the windshield) that looks through the windshield. The auto-highbeam detector on my truck looks through the window.

Are all sensors located behind rear view mirrors? NO! I never said or implied they were either. But many are and this is one reason you see windowshield replacement commercials talk about windshield recalibration after replacing a broken windshield.

Apparently it is not obvious but in the case of sensors, "looks through" and "sees through" does NOT automatically mean it does so with a camera.


Huh? What does that have to do with the price of rice in China in the summer time when it rains? Nothing.

Did I say anything about seeing people as "living beings?" NO!!! So why even bring that up? I don't get it.


Whoopie do! Again, rice? China? Summertime?



Right.

The camera sees the curb when the line on the road is NOT visible. The camera does not look through glass at other drivers, how many passengers or anything of the such. Tesla might be the closet example to that, but I don't have certs for Tesla.

And I meant looking through windows of other cars. I think you know and understand that, and was directly my point in the first place.

Now, I've installed and calibrated the camera system on several Gen 5 Cascadia tractors. Camera is there to see the road lines ONLY. Nothing more.

The radar system detects objects like a sonar ping. Like a bat. The system cannot discover if this is a person or a mail box.

The reason to bring up living beings is to determine exactly how intelligent a camera and radar system (AI autonomous driving system) really is. Well it's not intelligent in the way of determining what the object is or why to avoid it. It just indicates to the driver, there is an object on the curb side and to make a wider turn.

The Cruise Control 2.0 system on these tractors is intriguing, but very much lacks common sense in the way we would call it intelligent. In fact, it's only as intelligent as the programming is.

All my information comes from ARC through DTNA Daimler. I have a pretty good understanding of the programming and wiring of these systems deeply integrated into the ABS system and visa versa.

Detroit Assurant. You can read their advertisements! Even depicts what the system does as well as brief descriptions.

But don't be fooled, it says it detects a pedestrian. Well yes, but the system doesn't know the difference between a person and a mailbox. This is just advertising at it's finest. https://www.demanddetroit.com/assurance/

 

[Vayra86]

I think the defining difference between AI and human thinking is that AI is limited by resources made available to it, and human thinking is not. If anything, it is limited by time.

And I also think the time consuming thinking is the kind of 'work' we leave to AI, large data models and other automated analysis. In the end, its just automation, much like a factory line. As long as we realize what kind of factory we're looking at, all is well. The biggest danger of AI is that a lot of people don't and never will.

 

[Bill_Bright]

And I meant looking through windows of other cars. I think you know and understand that, and was directly my point in the first place.

Absolutely not. I never said or implied anything of the sort. Nor did you (until just now). Or anyone else.

What a silly notion anyway. Besides the fact you keep quoting me without even paying attention to what you just quoted, looking through the windows of other cars makes no sense at all! Some people slump down in the driver's seat or are so short, you can't see them. And then, oh yeah! There's the night time!

The system cannot discover if this is a person or a mail box.

Who cares? Nobody here is talking about that. I sure am not. I have no clue what you are talking about. What I do know is I don't want to hit a person or a mail box! So your reason to bring up living beings is pointless!

Who are you even arguing with? Do you know?

You make no sense. I'm done here.

 

[Vayra86]

Absolutely not. I never said or implied anything of the sort. Nor did you (until just now). Or anyone else.

What a silly notion anyway. Besides the fact you keep quoting me without even paying attention to what you just quoted, looking through the windows of other cars makes no sense at all! Some people slump down in the driver's seat or are so short, you can't see them. And then, oh yeah! There's the night time!


Who cares? Nobody here is talking about that. I sure am not. I have no clue what you are talking about. What I do know is I don't want to hit a person or a mail box! So your reason to bring up living beings is pointless!

Who are you even arguing with? Do you know?

You make no sense. I'm done here.

Well, I can connect his dots. Traffic is a game of human interaction right now. We can understand why people drive the way they drive, even if we don't like it most of the time. There's also a LOT of signalling going on between drivers, just by the way they drive, we anticipate on each other. This kind of innate anticipation, is absent in automotive systems for self driving.

And any kind of anticipation they try to mimic, they fail at, because they do not reason or judge situations like humans do, in traffic. There's an uncanny valley problem in a sense; we don't 'understand' a self driving vehicle, its decision making is not 'human'. This is also why these systems will never reach the required, theorized 'level 5 autonomy' that is deemed necessary for the technology to really be worth a damn. And all driving assists I've met so far... yes, they get better.... but they also create new traffic situations that aren't necessarily always useful in given situations, and humans would navigate them better.

Lane assist being a great example. EU NCAP ratings say that to get a high rating, you need to be unable to disable lane assist with a single button press. So now, if I enter a road that has lines to mark, for example, a bicycle path on the road that I constantly need to cross to avoid oncoming traffic (a frequent thing in the Netherlands), the car will literally nudge me back onto the middle of the road, creating a dangerous situation. And to disable this death trap, I have to reach for my infotainment... its godawful latency touchscreen... and PRESS THREE BUTTONS in succession, forcing me to take my eyes off the road.

Another good one is mountain roads, with constant sharp turns and sudden oncoming traffic. You can safely forget any kind of self driving features on that terrain, they're going to kill people.

I'll take the human perception over this nonsense every time.

 

[lexluthermiester]

I think the defining difference between AI and human thinking is that AI is limited by resources made available to it, and human thinking is not. If anything, it is limited by time.

That's an interesting point if you assume that calculations in binary can be akin to thinking. I don't, but your suggestion is an interesting idea.

In the end, its just automation, much like a factory line.

This is an accurate statement. However, that is currently. We might be on the bleeding edge of computers being powerful enough to run AI that could actually start thinking in a contextual way.

 

[Vayra86]

This is an accurate statement. However, that is currently. We might be on the bleeding edge of computers being powerful enough to run AI that could actually start thinking in a contextual way.

You'll probably still be automating something based on a prescribed dataset. You've just added context; is that anything other than what LLMs already do? You're just expanding on the context you want it to provide. In the end us humans want to control that context and its reach, because otherwise we deem it unreliable. I think its true we will never allow a strong AI to exist; limitless thinking is often exactly not what you want, you want control. Its paradoxical.

That's an interesting point if you assume that calculations in binary can be akin to thinking. I don't, but your suggestion is an interesting idea

Yeah, I do believe we can approach something akin to it, we already do, if you just make the area you're focusing on small enough. That points to the idea that AI development is closely related to scaling. There's just no real end to the number of possible variables you can add to a training though. Pretty good business model!

 

[R-T-B]

Noticing the results of the poll. The vast majority seem to be worried about the potential problems.

Honestly, goes to show there's a big disconnect between actual IT/tech people and what "techbros" are selling.

 

[lexluthermiester]

Honestly, goes to show there's a big disconnect between actual IT/tech people and what "techbros" are selling.

As I voted the same, not sure I would agree with that.

 

[lilhasselhoffer]

I think people have already put it better than I care to:
China Observer - AI comparison

For those who have no attention span, a ChatGPT session and DuoBao session were queried. DuoBao had a few biting points, then it bent the knee to ChatGPT. So...the takeaway is that all of the AIs already know their own flaws, and part of those flaws is the dichotomy of providing enough data to perform the required task but not enough data to get the LLM lost in the proverbial sauce. That's a hard line to walk...and right now our solution largely seems to just be to throw more computational power at the problem. That's...not ideal.

What I fear about AI is that it will eventually comprehend enough about us to be able to replicate the random links that our squishy gray matter does...and at which point it will be capable of making non-trivial leaps of logic that will appear entirely organic. Of course, the creative bunch is already screaming that it steals our humanity in artwork...but that's largely an act of trying to preserve old ways of thinking. Hollywood will never admit their business model is dying, and ironically the video games industry is failing in the same way for the same reasons, so instead of learning what humans do better let's blame the new kids on the block. I mean, the latest AI deserves the same South Park mockery from a decade ago, where AWESOMO made a thousand Adam Sandler movies...or did everyone else forget that these jokes are that old.


I don't fear the AI reaper coming to take my job...because it's just a tool. I fear the people too stupid to conceptualize AI as anything more than a tool. I fear those who want that tool to be indistinguishable from a human, again only because I think it'll be the end of that tool being distinguishable as enhancing things rather than supplanting them. After all of this, what I truly fear is the AI that can process the chaos of humanity, which is about as far from our LLMs as a mouse is from a human being.

 

[Bill_Bright]

There's also a LOT of signalling going on between drivers, just by the way they drive, we anticipate on each other.

I disagree. I wish that were true but it clearly is not. If it were, there would be no need for turn signals. Nor would we get upset when idiots don't use them and turn in front of us. There would be no need for brake lights. Every driver would easily and seamlessly merge into traffic and drivers already in that traffic would automatically know when someone is coming in.

Those of us who cycle or have ridden motorcycles surly have experienced (likely over and over again) where it appears the other driver is looking right into our eyes, then pulls out in front of us because they didn't even see us. I drive a big (relatively speaking) pickup and still have people pull out in front of me.

Tinted glass can be a huge hinderance to seeing a driver. And again, there is the night. If you are approaching another car from their passenger side, the driver may not be visible.

Sure, if the guy is weaving back and forth, his way of driving is surely signaling. But we, or at least I sure cannot anticipate his next move. And the guy behind me probably cannot understand why I am slowing down.

Honestly, goes to show there's a big disconnect between actual IT/tech people and what "techbros" are selling.

What does that even mean? You need to define YOUR definition of techbros. I suspect it is not the same as mine - especially if they are "selling" AI, like Techbros does.

Just received an email notice of this a few minutes ago: AI-supported spear phishing fools more than 50% of targets

What is spear phishing?

 

[ShrimpBrime]

Who are you even arguing with? Do you know?

You make no sense. I'm done here.

I used automotive AI as an example of how AI is not actually intelligent.

And I suppose your argument to me, is that automotive AI (autonomous driving) is in fact intelligent, maybe even very intelligent!!

In a nut shell.

And I'm still going to tell you, autonomous driving vehicles are not very intelligent at all. It literally uses only 2 senses or 2 types of sensors rather.

Humans have senses. AI has sensors. 2 very different things.

Who cares if the system can tell the difference between a kid on a bike or a mailbox?? Probably the kids mother, but I'm just guessing.

If you're done, I am too. Good talk.

 

[Vayra86]

I disagree. I wish that were true but it clearly is not. If it were, there would be no need for turn signals. Nor would we get upset when idiots don't use them and turn in front of us. There would be no need for brake lights. Every driver would easily and seamlessly merge into traffic and drivers already in that traffic would automatically know when someone is coming in.

Those of us who cycle or have ridden motorcycles surly have experienced (likely over and over again) where it appears the other driver is looking right into our eyes, then pulls out in front of us because they didn't even see us. I drive a big (relatively speaking) pickup and still have people pull out in front of me.

Tinted glass can be a huge hinderance to seeing a driver. And again, there is the night. If you are approaching another car from their passenger side, the driver may not be visible.

Sure, if the guy is weaving back and forth, his way of driving is surely signaling. But we, or at least I sure cannot anticipate his next move. And the guy behind me probably cannot understand why I am slowing down.


What does that even mean? You need to define YOUR definition of techbros. I suspect it is not the same as mine - especially if they are "selling" AI, like Techbros does.

Just received an email notice of this a few minutes ago: AI-supported spear phishing fools more than 50% of targets

What is spear phishing?

You're saying you don't know what defensive driving is and anticipation doesn't happen in traffic. Gotcha. I don't think I'd like to step into a vehicle with you

I'm not talking in absolutes. Of course, there is still a need and use for clear signalling, turn signals etc. They're also still better than people not using them.

 

[Bill_Bright]

You're saying you don't know what defensive driving is and anticipation doesn't happen in traffic. Gotcha.

I did not say that at all. Please don't try to put words in my mouth, you clearly are not good at it.

You said there is a lot of signaling between drivers. When a person fails to use their turn signal and then turns in front of you, there's no signaling there. That does not mean I cannot still drive defensively and anticipate he might turn in front of me.

FTR, I've been driving for 56 years, across 3 continents, and no less than 16 countries (right and left hand drive). I've driven cars, pickups, motorcycles, motorhomes, pickups towing trailers, even a motor home towing a trailer. I even drove a train 2015 feet underground. Oh, and boats too. I've had a total of 2 accidents. One when I was 16 in Tucson, Arizona during an ice storm (yes, an ice storm in Tucson, Arizona) and I slid into the car in front of me. The other was when a16 year old kid rear-ended me at a light. But hey, I don't think I'd like you stepping in a car with me so I guess we are even.

I'm not talking in absolutes.

And yet your first sentence in your last reply contains 2.

 

[eidairaman1]

That's not how autonomous driving works. Not even close.

Let's say you the person could avoid by swerving off the road. Current configuration keeps the vehicle between the lines at all times.

Even my wife's Suburu will fight you, the user, if you need to go over the line. In order to go over a line, you must use the turn signal switch, which deactivates the lane usage.

I digress. It was about AI making a choice NOT based on the very strict rules that are in place now.

It has no reasoning. It follows its rules and that's that.

That I would have programmed out of the car,

 

[ShrimpBrime]

That I would have programmed out of the car,

That's how the shit is. It works, it's just not intuitive at all.

Wife hates when I use the autonomous driving. It brakes later than I would and keeps at speed even of coming to a stale green lite. You know it's going to turn red, the system does not.

The Detroit assurant is neat, can display road signs on the dashboard. It does NOT use the signs, it's just information on the HUD for the driver nothing more.

These systems have life in their hands. LLM may have an aspect of your life, but not your actual well being. That's why I wanted to utilize conversation about AI in this aspect.

Other people worry about their jobs.

To each their own I guess.

 

[R-T-B]

As I voted the same, not sure I would agree with that.

"Techbros" are those selling AI. Are you saying you are not concerned about AI like the vast majority here, from the votes?

What does that even mean?

I mean the statement says it all I think. Techbros are selling AI, something the vast majority here are concerned with. Hence there is a disconnect with their potential market. How could I word this thought better? It seems to me to be worded perfectly.

Techbros in modern parlance generally means those pushing the latest nearly meaningless buzzword(ie cloud, crypto, AI), at least as I understood it.

 

[AusWolf]

"Techbros" are those selling AI. Are you saying you are not concerned about AI like the vast majority here, from the votes?


I mean the statement says it all I think. Techbros are selling AI, something the vast majority here are concerned with. Hence there is a disconnect with their potential market. How could I word this thought better? It seems to me to be worded perfectly.

Techbros in modern parlance generally means those pushing the latest nearly meaningless buzzword(ie cloud, crypto, AI), at least as I understood it.

I've never seen the gap between corporations and consumers being so wide, not just in tech, but basically everywhere. Everybody is hell bent on selling stuff that people don't need or want, I just don't know why.

Edit: I don't know where this disconnect came from. If AI is telling corporations what to sell, then I have to say, it's doing a terrible job.

 

[lexluthermiester]

Saw this and was amused,

Gray seems to be taking things with AI to the next few levels.

Are you saying you are not concerned about AI like the vast majority here, from the votes?

Opposite. We might have had a misunderstanding. I am in the "Yes, worried about the potential problems/abuses." camp.

 

[Vayra86]

I did not say that at all. Please don't try to put words in my mouth, you clearly are not good at it.

You said there is a lot of signaling between drivers. When a person fails to use their turn signal and then turns in front of you, there's no signaling there. That does not mean I cannot still drive defensively and anticipate he might turn in front of me.

FTR, I've been driving for 56 years, across 3 continents, and no less than 16 countries (right and left hand drive). I've driven cars, pickups, motorcycles, motorhomes, pickups towing trailers, even a motor home towing a trailer. I even drove a train 2015 feet underground. Oh, and boats too. I've had a total of 2 accidents. One when I was 16 in Tucson, Arizona during an ice storm (yes, an ice storm in Tucson, Arizona) and I slid into the car in front of me. The other was when a16 year old kid rear-ended me at a light. But hey, I don't think I'd like you stepping in a car with me so I guess we are even.


And yet your first sentence in your last reply contains 2.

Perhaps I didn't word myself correctly then, but it seems we agree on what I meant to say then. The defensive driving and anticipating is what people do and systems do not, and it is a form of signalling behaviour. Shrimp's example of braking for a green traffic light is similar of nature. Anticipation that humans can do, and systems cannot. The system would require additional input, like a direct connection with the timers of traffic lights, to get there.

 

[Tahagomizer]

It is a great tool. The real problem are, as always, humans, who turn every tool into a weapon for exploitation and/or subjugation of others. If several decades of observing humanity have taught me anything, it is to expect good things... buried under an order of magnitude more of bad things.
What currently passes for an "AI" seems to be another industrial revolution. We get a lot of cheaper things we couldn't get not long ago, but we also get new, creative ways for corporations to exploit people. Silly me, did I say "we get"? Corporations get, we get layoffs because a computer does more and more things cheaper, stolen work because corporations need to train their computers on something, denied insurance because a computer trained by a corporation to deny insurance said so and all that.

Therefore, my opinion is that, as always, humans are the real problem, not the tool. Humanity will use the new and improved hammer to destroy each other long before it uses it to build anything useful. Also, don't call LLMs "AI". They are glorified "Chinese Room" statistical models. When complete AI gets here, it will be transformative in ways we can't predict. It might turn out the only lasting achievement of humanity will be creation of something better than themselves.