In the news lately is an early 2024 political advert that shows us a
possible future if we have four more years of Biden/Harris. The reason
that it is in the news a lot is because it was (apparently) created using
AI.

If the objection was that AI was used to create alleged footage of a past or current event that didn't happen, I would be on board with the criticism. As it is, the advert is depecting a possible future outcome, and is very clear about that (2024 isn't now). I have watched it multiple times and have come to the conclusion that there is nothing any different about it than if someone had sat down with Shotcut and clipped together a bunch of photos and soundbytes, and then added some narration that sounds like it came from a news program.

That really is *not* anything new. The only difference I can think of
would be the amount of time it would take AI to do the same vs. humans
using editing software that is readily available FOSS. I could do that in
a day or two, with most of the time taken in gathering the source material
and recording any voice-overs. If I was part of a political organization
that had resources to get that material together for me, it wouldn't even
take a day... probably hours.

So I don't understand the problem with this particular advert, especially
when some of the people who DO see a problem with it DON'T see a problem
with the White House editing out of (or INTO) the archives things said
and done during Presidential press events.

The only harm I see in this "possible future" advert is that there are so
many people, left and right, who are so disconnected from reality that they might think these clearly future events have actually already happened. I
am not sure how much harm that actually does considering whatever else these same folks are going to believe/do/etc., as they've proven they will
overreact to *anything*.

The fact that it was done with AI, in this case, seems irrelevant.


* SLMR 2.1a * Oooo, Better run, Mr. Wino!!!
--- SBBSecho 3.20-Linux
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (618:250/1)

Hello Mike,

07 May 23 08:28, you wrote to ALL:

The fact that it was done with AI, in this case, seems irrelevant.

AI is made by humans and therefore is always imperfect and most certainly biased. ChatGPT is highly left-leaning biased and has been proven to be so by many different people.

-- Sean

--- GoldED+/BSD 1.1.5-b20220504
* Origin: Outpost BBS * Johnson City, TN (618:618/1)

On 07 May 23 16:55:24, Sean Dennis said the following to Mike Powell:

The fact that it was done with AI, in this case, seems irrelevant.

AI is made by humans and therefore is always imperfect and most certainly biased. ChatGPT is highly left-leaning biased and has been proven to be so many different people.

I knew someone was pushing an agenda when it was asked simple questions about Covid or what is the difference between men and women.

AI... Pffft... I've yet to have someone explain the difference between this "AI" and really good clever programming.

I'll know AI is a real thing when Sarah Connor arrives to save us all.

Nick

--- Renegade vY2Ka2
* Origin: Joey, do you like movies about gladiators? (618:500/24)

Hello Nick,

07 May 23 18:52, you wrote to me:

I'll know AI is a real thing when Sarah Connor arrives to save us all.

I like to call it artificial inelegance.

-- Sean

--- GoldED+/BSD 1.1.5-b20220504
* Origin: Outpost BBS * Johnson City, TN (618:618/1)

The fact that it was done with AI, in this case, seems irrelevant.

AI is made by humans and therefore is always imperfect and most certainly biased. ChatGPT is highly left-leaning biased and has been proven to be so

y

many different people.

Indeed it has.

Write a poem about Donald Trump.

"I cannot write about a politically hot topic"

Write a poem about Joe Biden.

"Joe is good, Joe is great, etc."

Mike


* SLMR 2.1a * How do you tell when you're out of invisible ink?
--- SBBSecho 3.20-Linux
* Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (618:250/1)

Hello Sean!

** On Sunday 07.05.23 - 23:39, Sean Dennis wrote to Nick Andre:

I like to call it artificial inelegance.

+1 :D
--
../|ug

--- OpenXP 5.0.57
* Origin: (} Pointy McPointface (618:250/1.9)

Hello Mike,

08 May 23 15:45, you wrote to me:

Indeed it has.

Just read a story about how contractors being paid $15 an hour are doing the actual programming for ChatGPT. What a joke.

-- Sean

--- GoldED+/BSD 1.1.5-b20220504
* Origin: Outpost BBS * Johnson City, TN (618:618/1)

On 09 May 23 14:19:43, Sean Dennis said the following to Mike Powell:

Indeed it has.

Just read a story about how contractors being paid $15 an hour are doing th actual programming for ChatGPT. What a joke.

Also likely being told exactly "what" to program.

Speaking of Skynet, its debatable that Linda Hamilton is still hot.

She's just that right amount of batshit-crazy that I find oddly sexy.

Nick

--- Renegade vY2Ka2
* Origin: Joey, do you like movies about gladiators? (618:500/24)

Sean Dennis wrote to Mike Powell <=-

Just read a story about how contractors being paid $15 an hour are
doing the actual programming for ChatGPT. What a joke.

I read something similar and thought they were doing the training for
ChatGPT, feeding content to it.

There's a scene in the later seasons of "Silicon Valley" Where Jin Yang,
one of the residents of the house, is trying to get funding for an app
that can determine what kind of food you're taking a picture of. It's a
bit of a staged demo, as all it can do is tell you if food is "Hot Dog"
or "Not Hot Dog". That's all he'd trained the app on.

One of the housemates tries to get his Stanford CS class to take
pictures of other foods to train the app, and the class pulls an end
run, make their own app and get funded by a Stanford alum who's a VC
auditing the class.

They end up selling their app to Snapchat to detect dick pics, so all is
well.




... Imagine the music as a set of disconnected events
--- MultiMail/Win v0.52
* Origin: realitycheckBBS.org -- information is power. (618:300/16)

Re: Re: AI in Political Ads
By: Kurt Weiske to Sean Dennis on Wed May 10 2023 06:36 am

Just read a story about how contractors being paid $15 an hour are doing the actual programming for ChatGPT. What a joke.

I read something similar and thought they were doing the training for ChatGPT, feeding content to it.

Some friends of mine are in a firm which is developing an AI backed intrussion detection system for networks.

What I hear from them is that half the work they do is write half the checks that determine if some traffic pattern is anomalous or not because the neural network is so bad at it.

Another guy I know is working in pollution control and they have tasked his team with having an AI determinate if the water sample in a picture is polluted or not. The problem there is lots of pollutans you can't detect with the nacked eye so the chances of the computer detecting pollution from most images are slim.

I think neural networks and machine learning have lots of useful applications, but AI has become the new buzzword du jour and everybody is trying to include neural networks in their projects only because it is cool to have a neural network like the rest of the kids.

--
gopher://gopher.richardfalken.com/1/richardfalken
--- SBBSecho 3.20-Linux
* Origin: Palantir * palantirbbs.ddns.net * Pensacola, FL * (618:250/24)

Kurt Weiske wrote to Sean Dennis <=-

One of the housemates tries to get his Stanford CS class to take
pictures of other foods to train the app, and the class pulls an end
run, make their own app and get funded by a Stanford alum who's a VC auditing the class.

Sounds like how Facebook was started...

They end up selling their app to Snapchat to detect dick pics, so all
is well.

The world always needs good dick detectors!

-- Sean


... WinErr 006: Malicious error: Desqview found on drive.
--- MultiMail/Linux
* Origin: Outpost BBS * Johnson City, TN (618:618/1)