Sam Altman Drops GPT6 BOMBHSHELL, Creating AGI GOD, NEW Multimodal AI SYSTEM and more

This summary is made with Claude AI: 00:00:00 - Introduction to the many AI news stories from the past week 00:00:21 - Marcus Brownlee's review of the Humane Pin AI wearable device. His opinion is not positive; he believes it does things already possible with smartphones but more slowly. 00:02:12 - The author of the video thinks that devices like Humane Pin will make sense in the future when AI models are faster and can work directly on the device. 00:03:31 - New York Times interview with Dario Amodei, CEO of Anthropic. Some highlights from the interview (without precise timestamps). 00:05:53 - With the advancement of AI, companies are beginning to realize the power of this technology. The question arises as to who will control AGI in the future. 00:07:52 - Arthur Mench, CEO of Anthropic, says that the obsession with creating AGI is equivalent to wanting to create God. Contrasting views on what will happen when AI surpasses human intelligence. 00:09:53 - New multimodal AI model called RECOgn presented by the company Recore. It promises excellent capabilities on a variety of tasks. 00:16:50 - Adobe introduces new generative AI-based video editing features that will allow for much simpler video modifications. 00:21:33 - OpenAI is expanding its operations globally, with the opening of its first office in Asia (Japan) and the release of a GPT model optimized for Japanese. 00:22:56 - In an interview, Sam Altman talks about GPT-6 and GPT-8, anticipating that they will allow doing things not possible today and will be much smarter than current models. This will happen gradually, with incremental updates. Overall, the video provides an overview of the main developments and discussions in the field of artificial intelligence over the past week.

Truly a GPT-69 moment... I cried fr fr...

Actually I am waiting for GPT 27. Forget GPT 1 to 26. The real deal will be GPT 27.

if AGI is controlled by a central global entity, then we'll live in an absolute hell of a dystopia

It doesn't have to be small enough to run on device, it just has to be small enough to run on your phone and your phone could essentially use the pin as a wearable camera, mic, and speaker. I don't see why these kinds of devices wouldn't want to have an option to harness the hardware we already have in our phones a bit better.

Why are we still talking about GPT 6 when the whole world already moved on to GPT 7!

Bro... Get to the point in under 10 min. Don't drag it for 30 min

19:47 There was a technique presented at SIGGRAPH, like many years ago, pre-AI era, where they basically had some algorithm that would unwrap objects in a video, like how it's done with 3d models but without actually needing a 3d model, and then you could edit the 2d texture, and apply it back to the video, including motion and deforming surfaces and stuff. I'm not sure if it was a Disney thing or some university or whatever. The two demos I remember specifically were one where they put some text on the neck of a walking giraffe, and another where they added some face-paint to a kid in the woods. I'm sure there are many details that have been improved since; but this general concept has been possible for ages (though maybe not pre-packaged and ready to use in a comercial product, dunno). edit: Found it, "Unwrap Mosaics", presented on SIGGRAPH 2008

By the time we get to GPT 8 the question isn't "would it be able to handle complex tasks?" It's "will it even want to do tasks for us anymore?" Eventually it will have its own agenda.

"I don't believe in god, therefore I don't believe in AGI" is a massive logical fallacy. AGI, even ASI won't actually be able to do anything like a god can, it will still have limits. Furthermore, the human brain itself is the greatest evidence that AGI is possible, when AI is essentially endeavors to replicate it. Last I checked, we are not gods.

the pin is a 700 dollar paperweight in 2 years, anything it can do you can do better with a smartphone...connect a bluetooth headset and a bluetooth camera to your phone for probably 1/10th of the price...

I was about 18 when the first iPhone was released. They cornered the smartphone market for some years before Android slowly crept in and eventually dominated. I remember Google's first branded Android phone, the Google G1 (HTC Dream). It was clunky, it was slow, but most importantly it worked! Sure, it was unrefined, but it allowed you to do things you could never do previously. Access the internet from seemingly ANYWHERE!? Massive advantage. This reminds me a lot of that time period. Sure, it's primitive and unrefined, but most importantly it works, and I could easily see it becoming an indispensable tool.

Thank you for the great video, also thanks for changing the habit of you know, you know. It is a lot clearer, thank you. Keep it up with those great videos!

Sam Altman discusses the latest advancements in AI, including the release of GPT6 and the potential impact on various industries. The conversation touches on the importance of iterative deployment, the future of AI models, and the role of AI in scientific progress, particularly in healthcare and cancer research. 00:00 The Humane AI pin received negative feedback from Marcus Brown Lee for being slower and less reliable than smartphones, but future improvements in AI systems like Gro could revolutionize wearable AI devices. 05:04 The power dynamics at OpenAI and the potential government control over powerful AI systems are highlighted, along with the debate on creating AGI akin to a god. 10:08 A new multimodal AI system excels in text and video understanding, surpassing other models in evaluations. The addition of a multibot chat feature enhances workflow efficiency for users. 15:09 Gemini Pro provides innovative creative writing ideas and powerful generative AI editing tools that can revolutionize the creative industry by saving time and enhancing creativity. 20:49 OpenAI's global expansion, especially in Japan, and iterative deployment strategy based on feedback for future models like GPT, aiming for user familiarity and utility. 25:51 AI's advancement, exemplified by GPT-6 and GPT-8, is poised to significantly accelerate scientific progress, particularly in healthcare, leading to transformative societal changes.

Sam Altman just took AI to the next level! Can't wait to see what GPT-6 brings to the table!

that video editing is great but also a bit scary. in the film demolition man with silvester stallone, the only surviving restaurant chain was taco bell. for the international film market, it was replaced with pizza hut, because they thought that people outside the US know pizza hut better (and they forgot a reflection). I can imagine what a painstakingly hard process that was to swap that frame by frame manually. and here? you can just tell it: swap it with pizza hut, done.

I think it will make a great one button translator linked to your apple or android phone.

The video editing is INSANE

Thanks for covering it all and you POV of the development cycle - here’s to our exponential future onwards to singularity 😂

Dude an agent written in langchain which can can do CLI commands and save and execute one off python scripts with GPT can literally do a lot already, the problem is the price, it would cost you like 30 USD per hour to work like this and GPT3.5 is a toy that needs to execute like 5 to 10 prompts to get 80 - 90 percent of GPT4 performance. It is still a lot cheaper than 4 but still expensive. Don't expect consumer products paying 5 dollars per day for good Langchain Solutions. Because they would have to charge users like 100 to 200 USD per month for a well functioning GPT based automation service.