Meanwhile tv manufacturers are working on AI to sneak in more advertising.

You could always just buy any TV with an an analog tuner and watch whatever's on the air these days.

Interestingly, in the novel Contact by Carl Sagan a rich character got his money by selling a device that did more or less that.

I just want a good quality TV without a smart os built in.

Unfortunately this does not financially benefit the tv manufacturers, and may land them in trouble with the platforms they themselves advertise on (like Google).

They're more likely to use AI to serve you more ads as an extra revenue stream; capitalism has gotta capital.

Not for most duff you mentioned, but the adbreaks themselves:

Our old dvr enabled us to skip ads in the recorded tv programs pretty accuratley. It set chapter markings whenever an ad-block began/ended which it figured out by the frequency of hard cuts as ads have them between every ad (so multiple times a minute) whlie normal programming usually does not. This was way pre-AI (like late 00s). Sadly the built in dvrs in our tvs after that did not have that function, but maybe there is a modern implimentation somewhere.

I agree but I don't watch TV so I don't bother. Yet... I still hate product placement so I might be interested in such a solution. Anyway here is how I would do it :

• evaluate what exists, e.g SponsorBlock, and see what's the closest that fit my need, try it, ask in forum or repository issues if modifications are possible
• gather videos of the typically problematic content, say few hours to start
• annotate them by adding the time stamps then the location on the image
• replace problematic content with gradually complex solutions, e.g black, average color of the area, denoising (quite compute intensive)
• honestly evaluate the result
• consider the biggest problem, e.g here on first pass fixed content so a detector based on machine learning for the type of content could help
• iterate, sharing my result back with the closest interested community

Honestly it's a worthwhile endeavor but be mindful it's an arm race. There are a LOT of smart people paid to add ads everywhere... but there are even more people, like you and I, eager to remove them. IMHO the key trick is, like SponsorBlock, to federate the efforts.

LibreELEC on a network with at least one Adguard home instance!!

"We're sorry, using AI-based ad-blockers is a violation of our Terms of Service Agreement. Per the agreement terms, your account is now suspended and you've been charged an additional early termination fee, because fuck you."

While I'm sure there will eventually be some grass-roots attempts, the providers will fight it to the death. A person can dream, though.

I just watch shows on a few apps with no ads. Britbox and Acorn are my go tos. They are premium but total about $15/mo. Also PBS app but there are a couple of ads as shows begin.

You cleanly have gotten this all wrong. The AI is suppose to make money not reduce income. Get out of here with this "make all ads completely vanish"

I run my tv through my pihole. It doesn't take out everything, but it's a good start.

I use npvr with comskip.exe and it does a fairly reasonable job of taking the ads out of free to air TV.

You can see in the timeline where it's detected ads, but you can use the mouse or arrow keys to still play those areas if it got it wrong.

Sorry, but the only AI TV you'll get has the job of analyzing your habits and selecting additional ads especially for you while completely trampling on your right or privacy.

I mean for advert breaks, there are projects to do this to recorded tv automatically (with varying degrees of success depending on the config and the channel).

That is, you record the TV from either a TV receiver card, or streamed live channels to disk, then run this process on the mkv/mp4/ts, and it will either create a set of chapters marking the ads (so you can skip them), or it will just remove them entirely.

I don't think it would transfer to "live" TV quite so readily though. Because it does scan the whole program to find things like logos etc to help work out where the adverts are. But, I mean a lot of the work has been done.

For removing all product logos. I mean, I bet we're not far from the processing power to make it possible. But, probably a fair bit of effort needed.

I can imagine the "AI" chips being neutered for these kind of tasks, like the "low hash rate" Nividia cards.

But I would love a TV that is smart enough to auto hide & mute every kind of ad. Even little logos on the athletes’ uniforms.

So, the best part about this example is that it is well intended but would have so many side effects it would be hilarious to see someone try to make it work. My assumption is that you want it to just have regular uniform colors where the ads are now.

The first assumption is that the team logo and colors aren't advertising. They are! Yeah, they make bank on tickets, but the real money is in merchandising. Merchandising only works because the people associate it with the team, so team uniforms at their core are ads. They weren't as much in the past when the majority of income was from tickets and concessions, but they are now. An easier version of this example is auto racing, where the car colors and entire paint job is an advertisement with a bunch of smaller ads plastered all over. Would the AI need to recolor all the cars to avoid color based advertising like bright yellow and black for DeWalt?

That also means that other media that exists to prop up sales in other areas are also ads. A lot of cartoons like Transformers, GI Joe, and My Little Pony existed as advertisements for the toys. The best way this gets convoluted is that Teenage Mutant Ninja Turtles (TMNT) was originally a comic book, which someone thought would be a great starting point for selling toys and to sell the toys they made a cartoon. But then they stopped selling the toys for a while, so the cartoon reruns weren't really ads at that point in time the same way they were originally. So does the TMNT cartoon always count as an ad because of the intent at the time it was created, or is it only an ad while the thing it is advertising is being sold?

Then you get into the fake ads in movies for things that don't exist. Are they ads? What about media where a real world thing is part of the plot, like how the military being in a movie is likely to be intended as an ad for the military?

I'm sure the idea is that the AI would know what the user means by ads, but the viewer will always be surprised when things they don't realize are ads get blocked and it would have to adapt to each individual viewer. Even more fun when multiple people try to watch something and they aren't on the same page about ads that impact the ability to watch!

I still love the post, but thinking how it could play out even if it worked is kind of funny.

A pet pieve of mine is people randomly sticking the term "AI" into a description of some particular tech solution.

You want ad blocking. (Which is based.) But you don't want "AI". If this can be done in a way that doesn't qualify as "AI", that would satisfy you, yes?

And using the term "AI" that way makes it clear you haven't really thought through what you really even want in that feature. (Not that there's anything particularly wrong with that, especially in a showerthoughts community, but it's still kindof a "slaps me in the face" kind of thing.)

And the term "AI" is so imprecise anyway.

And particular kinds of "AI" are such a bubble right now. And that's why everybody is sticking the word "AI" into random contexts for no fucking reason. But it's also just a gimmick at best and a huge scam at worst.

And "AI" is inevitably bad about false positives and such.

I'd really rather see the word "magic" than "AI" in this context. Because at least that admits that this is an idle wish and not something you think actual real-world adult humans should be seeking venture capital to attempt.

I'm sorry for taking this out on you specifically. You're definitely not the first person I've seen do this.