The AI apocalypse is here

#linguistics #ai #culture #polyglot #translation #chatgpt

[AWriterWandering]

The danger is not just in convincing people that something is real, but also in convincing them that nothing is real.

[Billyable]

This feels like the release of the internet 2.0. We're going through the "you can't believe everything you see online" phase again.

[frugal]

There was a recent scam where an employee got confirmation from his CEO via a video call that the payment was authorised, and ended up transferring several million dollars. The video call from his CEO was of course a deep fake...

[oldmanbanjo]

Yeah, I can hear it really clearly, but then I edit audio for a living. There's weird out of phase sounds, a lack of aspiration on vowels and way too much compression and saturation in order to hide the AI.

[JemRochelle]

I recently got a similar ad on Instagram for that language learning app and it freaked me out, the AI avatar looked like a corpse that had been reanimated 😂😭

[reo.incognito]

So, you are saying that YouTube (AI) polyglots will finally be able to fluently speak 50 languages? Nice!

[SingYourStory]

Excellent-so much useful information. Thanks for adding your humor to a terrifying subject.

[RosanneLoriMalowany]

You have amazing patience to dredge through all this stuff. Thanks for your precise investigation.

[DrSpooglemon]

Literally everything that was said about receiving content critically is relevant to non-AI generated content too. The important thing, no matter the nature of the content, is to consider all possible interpretations including that it is entirely false. Whether it is an actual human talking into a camera or an obvious bot voice over some stock footage. If it relates to politics and global affairs you have to ask yourself what is the bias of the source. And that includes from official news outlets. Everything is biased in to some extent and even when what you are hearing is literally true it can be presented in a way that makes you more likely to form a particular opinion of it. It's true of weird AI generated content on YouTube and it's true of the most mainstream news outlets.

[sowercookie]

It didn't suggest "the shit's out of the horse" which is my fave, poor showing Gippity...

[TheTraderGuy]

We need to have robust detection software, thereby eliminating the use of 90 percent of AI.

[nyuh]

this vid is sponsored by an ai service. definitely taking at least 3 grains of salt

[jakeaurod]

I read enough science fiction to have seen where this was going decades ago. I'm not sure what the earliest example was, but a machine intelligence teaches itself how to create a human voice and face and do impressions of other real humans in Heinlein's _The Moon is a Harsh Mistress_ that was published in 1966.

I've thought about starting a youtube channel, and even 10 years ago I was planning on incorporating a confusing and random background with lights and reflections that would make it computationally difficult for an AI to replicate, just so that people would know my video was unedited and unaltered. Of course, that would require me to do everything in one take. But I'm OK with people seeing and hearing me pause and refer to notes as if I was public speaking.

[AdaAnytime]

As a language learning nerd, I’m excited about how GenAI helps language learners. At the same time, I'm troubled by the use of deep fakes for misinformation. I appreciate this video for its common sense approach. The media literacy checklist is helpful for any media we consume, not only video content that may be artificial.

[alexdamman]

For what it's worth, I trust nothing any longer.

Addressed to the REAL Language Jones: keep doing what you are doing.

[lmodfz]

Just to say that I do actually watch people's teeth/mouths because I find prolonged eye contact very uncomfortable and I've found that more often than not no one notices. Plus it helps me follow what they are saying a bit more, less easily distracted by what else is going on around.
So I would definitely notice anything odd going on with your teeth 😂

[danielbriggs]

"We gotta make sure not to let them replace our use of search engines." This reminds me of when I realized I gotta make sure to support my local Wal-Mart.

[travcollier]

The AI media apocalypse won't be some particular misinformation or such... That is happening of course, but what will destroy our ability to get reliable info will be the sheer qualtity of automatic generated stuff. It's pretty depressing, because as long as we have views/engagement = money, there's really nothing we can do to avoid it.

[BrentHurst]

Hello! I work at a public school. Would it be alright with you if we showed this to admins and/or faculty and/or students?

(Full disclosure: I'm in the South, so we probably won't be allowed to show it to students since you cursed at one point and because "Joe Rogan says wacky things" would technically be the school taking a side on a few political issues, but I could see us showing it among admins and possibly all the faculty.)

[davidintokyo]

Inversely, there was recently a contest for the best AI-generated photograph. The image that the judges awarded the best was later disqualified because it was discovered (drum roll) that the image itself was just a normal photograph. (Could be apochraphal, of course.).
But the important point is that the underlying technology can't "understand" language, can't relate linquistic statements to actual things in the real world, can't do any reasoning whatsoever.
It's a statistical word guessing game based solely on word patterns and frequencies.
As a 1970s/1980s generation AI person, I'm aghast at the stupidity of the current underlying technologies...