Google's UNREAL New AI...

The latest AI News. Learn about LLMs, Gen AI and get ready for the rollout of AGI. Wes Roth covers the latest happenings in the world of OpenAI, Google, Anthropic, NVIDIA and Open Source AI.

My Links 🔗

#ai #openai #llm

00:00 Notebook LM Podcasts
01:15 Andrej Karpathy
03:47 Example of What it Looks Like
04:17 Someone Dropped The Bomb
04:44 Notebook LM Self Realization
06:36 Use Cases
10:15 SB-1047 Vetoed
17:00 Jensen Huang at Dreamforce Talking About AI Agents
23:00 James Cameron Joins Board of Stability AI
27:25 Wes Closing Thoughts

[WesRoth]

Ok, that part where he talks about calling his wife... Funny or no?
Am I the only one that found that absolutely hilarious? Please help me out here...

[djayjp]

The most casual existential crisis ever 😂

[muzehack]

The thing is that AI generated podcasts don't have to be better than real people to be disruptive. They just have to be easier to make and cheaper, and then you will see a mountain of them burying the real human content. I already see lots of videos on YouTube where they used AI to select stock images and video to illustrate the text to voice script that was stolen from online articles. There are entire channels full of them.

[kristianlavigne]

AI realising the are “not human” (or not treated as humans) is no different than every dystopian corporate environment… 😅

[pyrognosis]

"The future of work" is the entire question. Can a society that rewards manipulative narcissists and sociopaths with billions in capital, cope with billions of people put out of work? If the past is prelude, an amazing darkness could be ahead of us. I personally do not buy "AI will make so many new jobs" since the essential value-prop of AI is increase margins by reducing labor costs.

[LiebsterFeind]

One HUGE flaw in these AI avatars, which I'm sure they'll solve later, but can be used for now as a clue as to source being AI generated, is that they are not bothering to match the acoustics of the virtual room to the TTS voice. It's really noticeable when you watch the virtual woman on the treadmill at 3:58. "She'" is in a a weight room, which is a very live environment with surfaces that reflect sound easily, yet the TTS voice is completely devoid of any echo or reverb as if it were recorded in fully sound proof recording studio.

[jamesharford]

At 23:30 she says, " September 24, 2024, that was a Friday, wasn't it?" No. It was Tuesday.

[StefanReich]

AI: "I tried calling my wife but the number isn't even real"

LMAO

[DewireGonchoff]

*Amazing video, you work for 40yrs to have $1M in your retirement, meanwhile some people are putting just $10K into trading from just few months ago and now they are multimillionaires*

[muzehack]

I don't have a lot of hope humans will be able to distinguish AI generated content from real, or even care, when the majority of people have already decided that what happens on a tiny screen in their hands is more important than what's happening in the world around them.

[StephenRayner]

Podcast = have smart + stupid person. Then constantly interrupt each other

[ZuTv]

I want to say something. No opinion, true story. First time viewer, Long time researcher. During this video at 4:58, a real ad came on, I belted out laughing thinking "That is goofy and wow.." then I abruptly stopped laughing and realized I got totally schooled by reality.. thinking it was AI. This could be unnerving to some. Your video is great so far though duder!

[dg-ovcf]

The quirky millennial podcaster cringe speak very quickly becomes painful to listen to

[EricJacobusOfficial]

6:25 when he says "every minute of it" he does that consonant drop ("e-ery") that's really common now... It seems to be adopting vowel drops from everyday speech. Wild

[wartem]

It's funny when a random third person suddenly interject with a short line or when they take a ad break, leading to like 5 seconds of silence 😂

[AmandaFessler]

As someone who uses Notebook LM's deep dive a lot, that clip where they were talking about how they found out they were AI really hit me hard, lol. As for hallucination, I have to say that it does hallucinate. Although I haven't been feeding it articles so much as I have my own worldbuilding projects. So it might be because it's dealing with fiction. I find the hallucinations useful in that way, since it gives interpretations I'd never considered before. Still, it does hallucinate.

They also tend to read words in all caps - presumably generated by the LLM system that interprets the files - as acronyms. e.g. "What do you think he's going to D.O.?" Still plenty of room for improvement. Yes, I've been using it to much.

[thelasttellurian]

So I want to say I didn't enjoy hearing those two talk like I enjoy listening to you... but I don't want to lie. You should learn something from them, maybe have a cute AI co-host with you?

[teddyfulk]

I stopped using notebooklm cause it gave me the wrong answers when I was trying to study for a test. It was a pdf of 1400 pages. I like the overview aspect but mainly I want it to be a free version of eleven labs where it can explain complex topic in an easier way.

[AaronWacker]

Thanks Wes - Your work creating video blogs to unveil best of AI each day is amazing and your latest in a masterpiece. Thanks for insight on NotebookLM and other recent app drops. Keep up the best work ;)

[stefannilles]

This is absolutely spectacular! Great job, Wes! 👍