The A.I. Thread

[Firebot]

It's worth worrying about, but there's honestly nothing you can do about it. Live AI deepfake streaming is the holy grail for scammers and phishing. It's coming faster then the world is ready for it, and no AI regulation will prevent it, as it will simply come from a non western source.

Even Open AI allows you to create your own cameo using your voice and likeness, or any character just from a video. There's money to be made in deepfakes.

All you can do is prepare for the inevitability and be ready for the AI deepfake armageddon. It's already very possible and already used in elaborate scams.

https://www.cnn.com/2024/02/04/asia/...-kong-intl-hnk

[Shazam]

Kiro is the real deal. Built an app from the ground up, actually set up the db, caching, front end, back end.

[edslunch]

Is there an app that would let me upload a video of an awkward open mic performance and turn it into a cool performance video? Yes, it’s a pathetic cringe-worthy ask.

Edit: the audio is fine, it’s the visuals

[Firebot]

Realistically Wan 2.2 Animate would be your best bet. You can use it as rebranded versions on sites such as Higgsfield (they call it Higgsfield Animate). You would need to cut up your video in parts, but what you would do is upload your video, add an image of what you want to replace the scene with (maybe a concert etc) of and all of a sudden you are performing at a concert.

You may have to do some post video editing work to get a solid video. Although most of these are tailored to changing the character, you can definitely play around and get a good result. This is really the best simple option unless you want to delve into ComfyUI insanity.

https://higgsfield.ai/animate

https://higgsfield.ai/app/recast

You can also try restyle through Openart

https://openart.ai/video/restyle

Other open you can have (results may vary), upload the video as a character in Sora 2 which is a brand new feature which clips a few seconds of your video to build from, and use it to build something from it with the same character (you can even add the lyrics)

There's a few tools that can get something good, just perhaps not necessarily what you are looking for.

[Firebot]

Quick update on my AI journey since I hit a pretty cool milestone over the weekend. On Friday I got a message from CapCut's business account. I was announced as one of 25 Audience Choice winners with a 3000$ USD grand prize along with a number of other perks for a recent contest. Considering how huge CapCut is and TikTok as a whole and the field I would have been against, it's a pretty incredible win and getting recognized on such a level.

This prize propels me to over 10k USD made since I started my journey (March 2023), and it's clearly only beginning with the momentum I built. Lots of other positives in development as well.

[Fuzz]

LeCun, chief AI scientist of “fundamental AI research” at Meta, is expected to resign from Meta soon according to multiple reports from credible outlets. LeCun is a 65-year-old elder statesman in the world of AI science, and he has had seemingly limitless resources at his disposal working as the big AI brain at one of the world’s largest tech companies.

He’s actually been hinting at the answer for a long time. When it comes to human-level intelligence, LeCun has become notorious lately for saying LLMs as we currently understand them are duds—no longer worth pursuing, no matter how much Big Tech scales them up. He said in April of last year that “an LLM is basically an off-ramp, a distraction, a dead end.” (The arch AI critic Gary Marcus has ripped into LeCun for “belligerently” defending LLMs from Marcus’ own critiques and then flip-flopping.)

“If I tell you ‘imagine a cube floating in the air in front of you. Okay now rotate this cube by 90 degrees around a vertical axis. What does it look like?’ It’s very easy for you to kind of have this mental model of a cube rotating.”

With very little effort, an LLM can write a dirty limerick about a hovering, rotating cube, sure, but it can’t really help you interact with one. LeCun avers that this is because of a difference between text data and data derived from processing the many parts of the world that aren’t text.

https://gizmodo.com/yann-lecun-world-models-2000685265


I think this is a pretty reasonable view of the reason why LLM's are a dead end to AGI. Not to say they are useless, but we can't get there from here, no matter how much money is shove led at it. This realization may eventually pop the current AI bubble. It's also why they are not a good match to computer driving. They can't visualize the real world.

[Russic]

I'd probably have that feeling too if I worked at Meta.

In all seriousness though, there is a lot of hype bubbling up for the next Google offering. I find Google bad for that (their demo videos especially), but who knows, maybe they've got something interesting.