Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 40 original podcasts with new episodes released every day.
The Coming AI Crisis
There is an absolutely hilarious video of “Joe Biden” clarifying his withdrawal from the presidential race circulating on the internet. A sanitized version runs below:
You can imagine what lies beneath the bleeps or seek it out on “X.” Because the video creator, someone who goes by the handle “Midnight Mitch,” decided to use PBS NewsHour branded video, the network felt compelled to issue the following statement:
A deepfake video of President Joe Biden bearing our logo is circulating on social media. Biden did not make this statement. PBS News did not authorize the use of this video and we do not condone altering news video or audio in any way that could mislead the audience.
Now, Mitch did the correct thing and watermarked his video. He’s not peddling it as real. A political consultant who did send out robocalls from “Biden” two days prior to the New Hampshire primary is facing a $6M fine and state felony charges.
Actress Scarlett Johansson is in a battle with OpenAI CEO Sam Altman after she accused him of stealing her voice. Celebrities have long fought against soundalikes and imitators. Bert Lahr, the actor who played the Cowardly Lion in The Wizard of Oz, was none too happy with cartoon character Snagglepuss, and when Kellogg’s used the character in a series of TV commercials in the 1960s, Lahr forced them to put a disclaimer on the screen that read, “Snagglepuss voice by Daws Butler.”
That’s not to say that this technology needs to be banned. Congresswoman Jennifer Wexton, a Democrat from Virginia, suffers from Progressive Supra-nuclear Palsy and it has robbed her of her voice. (She’s retiring in January.) Her staff fed past speeches into an A.I. program and the result is a very passable version of her. She uses it on the floor of the House and in committee meetings. For all intents and purposes, it is her speaking.
It’s one thing for movie studios to resurrect the dead with family permission (and often participation) but as home computers and software become more powerful, there will be amateur mischief makers out there creating a whole lot of trouble in the future.
Published in General
It’s obviously fake. Too coherent to be Biden.
It is certainly going to be harder to distinguish between the disinformation and misinformation our government sends out.
Yep. I laughed when seeing the original. Of course it was obvious that it was a fake, from the content and fluidity of speech. Biden might have sounded like this four years ago, but definitely not today.
That might be a key defense against AI, at least for us. They will be unwilling to make it sound as bad as the left actually sounds.
To save you the time of searching:
https://x.com/i/status/1815136459671626121
Priceless. Thanks EJ and Steve for posting it. This just gets better and better.
For now, at least.
It may be that we’re over-estimating the danger of “AI”, at least this wave of it. Funny as it is, even the unbleeped version doesn’t match the words and lip movements completely. You can tell it’s fake.
But its Achilles’ heel is its power and computing requirements. This generation of machine learning can only become so large, limited by the amount of coal and uranium we can mine.
The video is from his address following the assassination attempt. The PBS statement is inaccurate as it’s not technically “a deep fake” of image mapping someone else’s face over video. What’s new is the voice cloning. That ability has never been available before.
A.I. voice cloning is still in its infancy. There are unnatural inflections and strange pronunciations that often require phonetic spellings in text-to-speech models, especially with names. But it will get better.
Deleted.
I think it’s dead, Jim.
Rick Beato posted an interesting video about a month ago. He presented his kids with AI music and recorded music and didn’t tell them which was which. They identified the AI music every time. It shocked Rick, though, of course, he was happy about it. They insisted there was an electronic sound within the voices akin to AutoTune. Rick wondered if the sound was a frequency they were still capable of hearing that he couldn’t.
Social platforms would be within their bounds to ban AI-generated material, and it’s probably a smart thing to do given the increasing number of lawsuits that will be coming from people whose work is used to inform AI creations. People will whine about it, but many creators already operate with full knowledge of copyright restrictions. They’re used to it, and their original work may end up feeding the AI monster as well.
Funny! 🤣
I’m close to assuming everything is fake, and the faster we all get to that point, the faster we coalesce around trusted nodes.
another great fake out there-from a few weeks ago: