When is AI going to make everything else obsolete?

[dellsam34]

If I ask, when there will be a kitchen appliance that can make chicken salad out of chicken shit, I would be called crazy, and that would be correct. You can't make chicken salad out of chicken shit, In other words you can't make something that didn't exist in the first place. You want to get the best quality out of the tape first, then procede to minor corrections. You don't hire the cheapest tile installer, then hire another tile installer to fix the first tile installer's mistakes, You want to do it right from the get go.

Not specifically apply to the OP but this is a very common multi-stage syndrome I've been observing throughout the years of being in such forums. 1-Go acquire the crappiest capture hardware due to lack of knowledge or budget, 2-Being told the results are crappy, 3-Lookup online for better alternatives, no solution found, 4-Looking for magical solutions (Get together and invent a new capture card, invent a new TBC, AI alternatives), 4-Acceptance, looking for a better hardware to finally do it the right way, Or give up and never return to the forum, Some will have a meltdown and curse everyone here and leave forever.

[lordsmurf]

anyway, Really nice video !!

It's not his. Bwaak just links to random stuff.

Furthermore, that video gives standard terrible Youtube advice. All of the samples have various image damage as shown. All of the settings there are crap. If extremely low quality is your goal, then there you go. But his capture look worse than the data on the original tapes, and it didn't have to be that way.

[lordsmurf]

I could give you half a dozen links to YouTube videos right now that are stretched, have horrible combing, blown out highlights, not even mentioning broken motion, yet they have hundreds of thousands of views and thousands of likes.
Visit any sports bar and observe TVs showing stretched or squished picture, or showing an SD channel when HD is available, and no one cares.

I agree. However, the fact that people are like that is a totally different story. Actually, a sad one, if you ask me.

Not just sad, but often pathetic.

But comparing yourself to the masses is often a bad idea. Remember, a huge % are obese, and eat a diet of crap like McDonald's and ramen noodles. So would that lowest common denominator be your bar to good eating, and living a healthy lifestyle? No, of course not.

Folks in sports bars are often so drunk they don't know what's going on, who has the ball, etc. And the screens can be so tiny (and overhead) that it's mostly just background audio anyway. That said, I rarely see non-HD signals in any establishment anymore, unless it's some OTA channel with Gunsmoke/whatever re-runs (ie, doctor's office). Even the SD channels are broadcasting with black bars now, as the ads are 16x9.

[Bwaak]

https://www.youtube.com/watch?v=oYMszuMyGG4 anyway, Really nice video !!

What exactly is nice about it?

[plehoediv]

Confidence works both ways. I am as convinced that there will be, as you are that there will not be.

My statement is not based on belief: anyone can believe anything they want, but that belief won't sway anyone nor will it make something magically happen. Instead my statement is based on learning curve theory.

I understand.

With nuclear fusion, Lawrence Livermore Labs (and others) have been trying to create a net positive energy result from fusion for almost forty years. They finally achieved that goal for a small fraction of one second, but there is little in that experiment that would suggest they will be able to sustain that result for even a second, much less continuously. Therefore, given that it took forty years to get here, and given that there has been no new branch to the effort (like "cold fusion"), it will be at least 100 years before anything practical gets built.

I see your point, but...

"I think there is a world market for maybe five computers." - Thomas Watson, chairman of IBM, 1943

"X-rays will prove to be a hoax." - Lord Kelvin, President of the Royal Society, 1883

"Television won't last because people will soon get tired of staring at a plywood box every night." - -Darryl Zanuck, movie producer, 20th Century Fox, 1946

"If excessive smoking actually plays a role in the production of lung cancer, it seems to be a minor one." - -W.C. Heuper, National Cancer Institute, 1954

"The world potential market for copying machines is 5000 at most.” — IBM, to the eventual founders of Xerox, saying the photocopier had no market large enough to justify production, 1959

A rocket will never be able to leave the Earth’s atmosphere.” — New York Times, 1936

Say what you want to these statements, but I bet most of them are like yours, based on "learning curve theory".

Besides, many of those statements come from people that do not want a technological breakthrough because it would not benefit them. I do not know much about nuclear fusion, but there are more factors involved than just scientist sitting there trying to make something work. Thinks like politics and greed, for example. Anyhow, this is a video forum so there is no need to get too far of the tracks.

...while other more traditional software approaches have given us stabilization, dirt removal, denoising, sharpening, and a host of other technologies that can make fantastic improvements and are used everyday by lots of people.

Alright, after reading your last sentence I realized I made a mistake! Instead of using the term "AI", I should have said SOFTWARE, because that is what I really meant! I don't care if it is AI or Billy Bob who makes it happen. . I guess I used the term AI because I mixed up autonomous with AI, if that makes sense? English is not my native language.

Having said that, I am still hoping that in 10 years' time I can just throw "bad" footage at a Software and with the push of a button, perfectly fine footage comes out the other end.

Thanks.

[dellsam34]

Having said that, I am still hoping that in 10 years' time I can just throw "bad" footage at a Software and with the push of a button, perfectly fine footage comes out the other end.

No, That would never happen. Prediction is a guess of the unknown, Something is missing and want to replace it with a guess works only to a certain degree, Again as previously said, there is nothing about enhancing video hasn't been done already, AI does it in an automated way, that's the difference. What you think will be possible 10 years from now you can do it right now with a little bit of a know how and patience, But remember garbage in will always be "enhanced" garbage out. So you have a better chance at getting good results from the tape with better hardware right now before they degrade over time.

[plehoediv]

If I ask, when there will be a kitchen appliance that can make chicken salad out of chicken shit, I would be called crazy, and that would be correct. You can't make chicken salad out of chicken shit, In other words you can't make something that didn't exist in the first place. You want to get the best quality out of the tape first, then procede to minor corrections. You don't hire the cheapest tile installer, then hire another tile installer to fix the first tile installer's mistakes, You want to do it right from the get go.

Well, In theory, it is possible to analyze the components of poop, such as its chemical composition and microorganisms, and potentially use this information to create synthetic versions of some of its components. However, reverse engineering poop is a complex and challenging process due to the diversity of substances and microorganisms it contains.

Poop is composed of various organic and inorganic compounds, including water, protein, carbohydrates, fats, and minerals, as well as a complex microbial community. Each person's poop is unique and can vary depending on their diet, lifestyle, and health status, making it difficult to generalize and standardize the analysis process.

While it is possible to extract and purify some of the components of poop, such as enzymes and bacteria, and use them for various applications, such as wastewater treatment or bioremediation, the process of reverse engineering poop as a whole is not practical or efficient for most purposes.

Overall, while it is theoretically possible to reverse engineer poop, the complexity and variability of its composition and microbial community make it a challenging and inefficient process for most practical applications.

Thanks ChatGPT. I was going to say somewhat the same, ha ha.

Not specifically apply to the OP but this is a very common multi-stage syndrome I've been observing throughout the years of being in such forums. 1-Go acquire the crappiest capture hardware due to lack of knowledge or budget

,

Done that.

2-Being told the results are crappy,

Yup!

3-Lookup online for better alternatives, no solution found,

Yes and No. I am looking what to buy next. Probably a JVC VCR with TBC, because their TBC is supposed to be "more" of a TBC than other VCR's TBC. Do not quote me on that. I remember reading that somewhere. . Then get a capture card and maybe a second "real" TBC. So I did find a solution. I guess.

4-Looking for magical solutions (Get together and invent a new capture card, invent a new TBC, AI alternatives),

I would say this is usually number 1 with me. I always look for a magical solution first, then I go acquire the crappiest capture hardware due to lack of knowledge or budget. (Half true as budget is not an issue).

4-Acceptance, looking for a better hardware to finally do it the right way,

Like I said, doing that right now, ha ha.

Or give up and never return to the forum,

To 99.99% I do not give up on a forum. The only time I ever "gave up" was on a HAM Radio forum. LOL.

Some will have a meltdown and curse everyone here and leave forever.

That what happens to me all the time, but the other way around. They curse at me and hope I go away forever, ha ha.


Thanks.

[lingyi]

The limitation of AI and all computing is the programming/programmers. GIGO

AI colorizing can predict that sky is blue, grass is green. But it takes a human eyes, mind and experience to accurately determine what shades and other nuances to apply to the programming. And even then, it's limited to what is programmed to do. If I only program "blueberries are blue" and "blueberries are fruit", and "bananas are fruit, the computer would output "bananas are blue". I would have to add "bananas are green/yellow/red/pink/purple/black" and "bananas are fruit" and add the parameters of the different color bananas for the computer to output, "This banana is green/yellow/red/pink/purple or black", "But not blue"

Edit: Then, if someone genetically modified a banana to have blue skin, that information would have to be programmed in the code.

AI can never make human knowledge, imagination and intuition obsolete. A computer can build on Einstein's theories by compiling the collective theories past, present and future (by programming), but it can't know what it doesn't know (i.e. programmed).*

Edit: I would argue that AI couldn't define and classify a platypus if it wasn't programmed.

Computer

warm blooded = mammalian

egg laying = oviparous

webbed feet = avian

bill = avian

broad flat tail = beaver

swims underwater = amphibian

classify? = DOES NOT COMPUTE! DANGER! DANGER! *Magic smoke!*

*Edit 2: To expand and clarify. Yes, AI does make predictive assumptions, but it's not empirical knowledge in the human sense. Back to the sky and grass. AI predictively colorize the sky and grass, but it takes a human's eyes and mind to determine if it's empirically correct.

[Cornucopia]

Too add to that, sky and grass...I can think of hundreds if not thousands of varying shades of blue, green, gray, brown, purple, red, yellow, etc that I have already witnessed in different skies and grasses, sometimes in different sections of the same scene of a single instance. I know how things are colorized currently, both manual and automatic, even in the most advanced methods, and ALL of them pale in comparison to reality. It would be a shame if those methods (or even improved versions) became the norm and not the special exception*, and we were left with plastic representations of reality. No thanks!

(*Note: it is ok for a few exceptions)

Scott

[lordsmurf]

AI colorizing can predict that sky is blue, grass is green. But it takes a human eyes, mind and experience to accurately determine what shades and other nuances to apply to the programming. And even then, it's limited to what is programmed to do. If I only program "blueberries are blue" and "blueberries are fruit", and "bananas are fruit, the computer would output "bananas are blue". I would have to add "bananas are green/yellow/red/pink/purple/black" and "bananas are fruit" and add the parameters of the different color bananas for the computer to output, "This banana is green/yellow/red/pink/purple or black", "But not blue"

You stole my favorite example.

John likes bananas.
John likes blue.
Bananas must be blue.

[lingyi]

Smurfs are blue
lordsmurf is blue
All lords must be blue!

Therefore:

WE should all be blue, like lordsmurf too!
<3 <3 <3

[lingyi]

On AVSforum, there was a poster who claimed he calibrated his TV by shooting the greenery outside his window and matching the color, tone and contrast to that. He was quickly torn about by those pointing out his camera gave a limited spectrum of reality and a second later, the sunlight would minutely change the values of what he recorded. And his "calibration" didn't apply to any other other scene than that particular video. Again, AI is derived from limited data = limited programming = unrealistic extrapolations.

The other day, I told someone the story about how I was infatuated with a co-worker whom I only saw in the limited spectrum of the office lighting. That infatuation ended when I saw here in bright daylight and every minute flaw in her face that was hidden by the office light was revealed!

AI: Is that a mole, a pimple, a bug, a dirt spot, a spot on the lens, a digital artifact?

On a thread about logo removal, a poster claimed that AI could accurately recreate whatever's under the logo. I posted a screencap of someone's face obscured by the huge station logo in the top right corner. Yes, a filter could extrapolate what the person's face is like under the logo based on before and after shots, but there's no way unless their was access non-logo source, could the [AI] exact facial expressions made [by her and be] recreated.

[Hoser Rob]

Scott and John Meyer. Both very smart persons.
I studied AI 60 years ago and learned then that it would be a very long time for AI perfection.
Something approaching a human brain may someday succeed.

Not even an octopus brain in the foreseeable future.
And "artificial intelligence" is a huge misnomer, to begin with.

Thanks for some sense. Generative AI is just massively, riidiculously overhyped right now. Esp. since the tech sector is being so hammered in the markets right now. They're desperate for buzz.

[sophisticles]

AI, lol.

It reminds me of the whole "Cloud", "it's in the Cloud", "Cloud Computing".

Do you know what the "Cloud" is? A bunch of remote servers, that's all.

Want to know what "AI" is?

A Bunch of carefully structured if/else/elseif or better yet a proper constructed case/switch block.

That's all "artificial intelligence" is, a programmer that puts together a bunch of conditional statements designed to account for every possible input.

If you have even written anything like this:

take some input
make a bunch of decisions
return a reply based on the decisions made

Congratulations, you just created AI.

[TreeTops]

sophisticles...if that is true, and I believe it is, then AI hasn't progressed much at all in the last 60 years. What you said above is exactly what my AI instructor taught me 60 years ago.

[sophisticles]

sophisticles...if that is true, and I believe it is, then AI hasn't progressed much at all in the last 60 years. What you said above is exactly what my AI instructor taught me 60 years ago.

If you think about it this is how rational decision making is done, information is acquired, then it's analyzed, then a course of action is decided upon.

The only difference is that artificial intelligence is a term for machines.

If you add a random element to a series of conditional statements, and some historical data that acts as "experience", then you can mimic natural intelligence.

[dellsam34]

The difference is computers' speed and storage increased which allowed new and sophisticated models to be made, The concept is still the same.

[El Heggunte]

Through my Yandex email account...

Microsoft tries to justify A.I.‘s tendency to give wrong answers by saying they’re ‘usefully wrong’.

KEY POINTS

*Microsoft debuted ChatGPT-like features dubbed Copilot that will eventually come to its business apps like Word and Excel.

*Current AI technologies like ChatGPT can often produce inaccurate responses, and some technologists are fretting over the possibility that people may place too much trust in the software.

*Microsoft is pitching the technology as being “usefully wrong.” It’s a spin on AI’s tendency to occasionally stumble with facts.

https://www.cnbc.com/2023/03/16/microsoft-justifies-ais-usefully-wrong-answers.html

[plehoediv]

The limitation of AI and all computing is the programming/programmers. GIGO

AI colorizing can predict that sky is blue, grass is green. But it takes a human eyes, mind and experience to accurately determine what shades and other nuances to apply to the programming. And even then, it's limited to what is programmed to do. If I only program "blueberries are blue" and "blueberries are fruit", and "bananas are fruit, the computer would output "bananas are blue". I would have to add "bananas are green/yellow/red/pink/purple/black" and "bananas are fruit" and add the parameters of the different color bananas for the computer to output, "This banana is green/yellow/red/pink/purple or black", "But not blue"

Edit: Then, if someone genetically modified a banana to have blue skin, that information would have to be programmed in the code.

AI can never make human knowledge, imagination and intuition obsolete. A computer can build on Einstein's theories by compiling the collective theories past, present and future (by programming), but it can't know what it doesn't know (i.e. programmed).*

Edit: I would argue that AI couldn't define and classify a platypus if it wasn't programmed.

Computer

warm blooded = mammalian

egg laying = oviparous

webbed feet = avian

bill = avian

broad flat tail = beaver

swims underwater = amphibian

classify? = DOES NOT COMPUTE! DANGER! DANGER! *Magic smoke!*

*Edit 2: To expand and clarify. Yes, AI does make predictive assumptions, but it's not empirical knowledge in the human sense. Back to the sky and grass. AI predictively colorize the sky and grass, but it takes a human's eyes and mind to determine if it's empirically correct.

A brain is like a computer and a computer is like a brain. Both use electrical signals to send messages. The brain uses chemicals to transmit information; the computer uses electricity. Even though electrical signals travel at high speeds in the nervous system, they travel even faster through the wires in a computer. Both transmit information.

So, if one would be "limited by the programming/programmers", so would be the other. In our case, the programming/programmers are the People and things around us. You wouldn't make a statement like you did (at the beginning) about humans, or would you? Just because our parents, family, friends and surroundings are "limited", doesn't mean we will be and that we cannot surpass those limitations.

Besides, the goal of artificial intelligence (AI) is to create intelligent machines that can perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, language translation, and many more.

The overall objective of AI is to develop intelligent agents that can learn from experience, reason about the world, and take actions to achieve specific goals in various domains. This can lead to improvements in a wide range of fields, including healthcare, transportation, finance, education, and many others. Ultimately, the goal of AI is to enhance human capabilities and improve the quality of life for people around the world.

Thanks.

[El Heggunte]

.........
The overall objective of AI is to develop intelligent agents that can learn from experience, reason about the world, and take actions to achieve specific goals in various domains. This can lead to improvements in a wide range of fields, including healthcare, transportation, finance, education, and many others. Ultimately, the goal of AI is to enhance human capabilities and improve the quality of life for people around the world.

You are wrong.

[Cornucopia]

Ignoring all the BS of hype and anti-hype, it boils down to this:

AI uses iterative processes based on given "models" representing intended archetypes of the "real world". AI is only as good as -
1. The # and type of models (which is limited by storage, memory, previous model generations).
2. The choice of archetypes of the real world (which is limited by programmer decisions on what is a good archetype).
3. The iterative decsions followed (which is limited by programmer decisions on what is an appropriate method of coming to a result for a given series of models).
...maybe more constraints.

Given those constraints, its easy to see that AI is not going to be a major replacement, or improvement, on human processes anytime soon. They WILL grow in use as helpful TOOLS (once better understanding of the pitfalls of blindly relying on them can be incorporated into the process).

As an example, I have been spending a lot of my COVID and post?-COVID time learning more languages (currently besides English, learning German, Italian, Hawaiian, Cornish, Korean, with expectations for about 3 or 4 more), and because of my AV & IT background, have been interested in following the state of the art of machine translation.
It has greatly improved over the last few years, but it still is in its toddlerhood, if no longer infancy. It works very well with "well-formed", grammatical, text-only input of certain standard prose styles using a limited palette of source and destination languages. But, if you give it REAL WORLD input of children speaking, people with heavy accents, or poor grammar, or music lyrics/poetry or legalese, or want to start or end in less popular languages, you will usually end up with either partial or full GARBAGE.

I think many give computers more credit than they are due, and humans less credit than they are due. Certain people excepted.

Scott

[plehoediv]

.........
The overall objective of AI is to develop intelligent agents that can learn from experience, reason about the world, and take actions to achieve specific goals in various domains. This can lead to improvements in a wide range of fields, including healthcare, transportation, finance, education, and many others. Ultimately, the goal of AI is to enhance human capabilities and improve the quality of life for people around the world.

You are wrong.

ChatGPT wrote that for me.

[Cornucopia]

Yeah, you can tell. "Increase the quality of life..." sounds like every corporate-speak boilerplate BS I've heard.

Scott

[lordsmurf]

I think many give computers more credit than they are due,

I've said this for decades. A computer is an electronic hammer, screwdriver, lawnmower. Just a dumb tool that "smartly" does a few things. And "smartly" means it makes our life easier.

These are all just mild tool upgrades, not some game changer.

- Hammers with laser sights for precision.
- Square screws (and drivers) that don't slip or strip, vs. flat head.
- Self-propelled mowers, vs. push. Also better than riding for tight conditions.
- Computer software that attempts to anticipate needs vs. always responding. But noting it can easily be as wrong as we are, their crystal ball sucks too.

[lingyi]

Program everything Shakespeare wrote and every Olde English dictionary and you'll never get a new play that is definitively what Shakespeare would have written. This the reality of AI. It can imitate style, but it can't create something from what isn't there.

Okay. I'm out. There will never anything said that will convince the OP otherwise. We'll continue to let him/her believe and those of us who understand the limitations and truth of AI will continue to know.

[rrats]

I have been playing with ChatGPT for a while. Funny how often times I tend to get confidentially incorrect answers

[_Al_]

If AI was participating here, it would at least stay on topic.

[plehoediv]

Okay. I'm out. There will never anything said that will convince the OP otherwise. We'll continue to let him/her believe and those of us who understand the limitations and truth of AI will continue to know.

Why would you say that? No one needs to "convince" anyone of anything. That fact that you would say such a thing shows your "limitations", to be honest.

EDIT: I mentioned earlier (in this thread) that I shouldn't have used the term AI. I should've said software, with the help of AI instead. Maybe. Why? Because I actually do not care if it is an app you install on your PC, a cloud-based app, a software, a software with AI, a plugin or a special little box. I do not care.

That said. People keep saying that if there isn't enough information in the original footage, software/Ai or whatever you want to call it, will not be able to fill in the "gaps". I say it will. Think about those car commercials where they use a battery-powered automotive rig with a fully-adjustable wheelbase, track width, and suspension setup to mirror any car. The rig provides video producers with integral raw footage—like how a car dynamically behaves.

Same should [one day] work with old footage. The software looks at the footage of let's say your old car and uses that to get the integral raw footage it needs. The program then downloads the dimensions of your car and basically recreates it again on Top of the old footage. Just like they do in car commercials. Honestly, isn't that what they are doing with deep fakes?

Thanks.

[_Al_]

Honestly, isn't that what they are doing with deep fakes

How does one sets boundary, dividing line? Does it matter. Do we want it?

Imagine a slider in "AI enhancing software". You start to move enhancing slider from zero to the right. Right after zero you are enhancing only a tiny stuff, more like cleaning a footage, slight denoise and slider indicator says "cleaning". You keep moving slider to the right, some other filters are added on, you can read "enhancing". But at what point it starts to read: "deep fake"? Who is to say so. If you have two pixels and software generates a hair, it is just fake hair, one of many possibilities of the original footage. Not only hair, but it can fake generate a complete face adding features, that face after enhancing was not that face.

Just mere cleaning introduces a mistakes and faults. There is a human comparing images and says, that filter is too much, or not enough. But mostly that implicates "lost" details, not added.
Funny stuff starts to happen in opposite direction, for example, maybe QTGMC is the filer that starts to add features and therefore, it is already in a category "fake" even if not that obvious? Is any deinterlace already in that category?
Point is how much , where is a border from enhancing a footage to reconstructing it and therefore having a simple fake. Who says where is that border? Who is to say that is fine but from here it is fake. This is a mind blowing, "gods boundary" whatever can be called and no one is talking about it, at least here, maybe it is a topic somewhere.

[usually_quiet]

....and because of my AV & IT background, have been interested in following the state of the art of machine translation.
It has greatly improved over the last few years, but it still is in its toddlerhood, if no longer infancy. It works very well with "well-formed", grammatical, text-only input of certain standard prose styles using a limited palette of source and destination languages. But, if you give it REAL WORLD input of children speaking, people with heavy accents, or poor grammar, or music lyrics/poetry or legalese, or want to start or end in less popular languages, you will usually end up with either partial or full GARBAGE.

I think many give computers more credit than they are due, and humans less credit than they are due. Certain people excepted.

Scott

Very smart people have been working doggedly at machine translation and natural language processing for more than 40 years to get that far. (I met someone working in that area about 40 years ago.)