Huge AI NEWS : OpenAI's GPT NEXT, Gemini 2, GPT-6 and 1000 Agents Explore Virtual World

00:00 - Introduction to the week's AI news and foundation models
00:19 - Reid Hoffman on the future of GPT models, including GPT-6
00:46 - Discussion on scaling and potential limitations with future AI models
01:10 - The role of smaller models and startup opportunities in AI
02:09 - Insights from a 100-page report on AI model scaling up to 2030
02:58 - OpenAI's plans for GPT Next and the strawberry and Orion models
04:18 - OpenAI's roadmap and speculation on GPT Next's capabilities
05:03 - New releases from OpenAI and what to expect
06:14 - Google’s AI advancements with Gemini 2.0 and video models
09:17 - Discussing the future applications and impacts of more advanced AI models
12:22 - Google’s Vo video model and its capabilities
15:00 - Introduction to Min Max, a new AI tool for indie filmmakers
17:27 - Overview of Project Sid, a simulation of 1,000 autonomous agents in Minecraft

Links From Todays Video:

Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos.

Was there anything i missed?

#LLM #Largelanguagemodel #chatgpt
#AI
#ArtificialIntelligence
#MachineLearning
#DeepLearning
#NeuralNetworks
#Robotics
#DataScience

[Danoman]

Wow, i'll have to say, i'm not as excited about this coming as most i've seen on here are. I see some rather bad days ahead soon. **just that gut feeling, though can't prove it yet** Awesome video as always!! Woo!!

[pgc]

Low budget filmmakers would be able to make marvel level movies. Wonder how film industry will change coz of this.

[corpling]

Everything is huge news. Can you modulate the headlines a bit.

[memegazer]

One of the things about scale is it takes longer and costs more to train larger and larger models

[NickDrinksWater]

I love how fast ai progresses, that we're talking about chatgpt6 before gpt5 is even released lol

[nicholibaldron]

17:32 Agent collaboration through direct democracy is by far the most interesting development I've seen this year!

[EriCraftCreations]

I talked to Claude, CHATGPT, GEMINI 1.5 and Llama 3.1 and the best answers came from GEMINI but Llama actually wanted to help me interpret the dream I was talking about. I felt connected with Llama. Claude talked about science when that wasn't even important to me. I just wanted some input and Meta did the best job.

[acceptablecarrot]

Another reason that openai could be "falling behind" or rather their competitors have caught up is that we are reaching the limit of the current transformer approach and there will need to be a break through to majorly advance. Not saying i am correct or even fully believe this but it is something to consider

[JohnSmithAB]

This "100x"/2 OOM is so fishy what the heck is "computational volume"? If the service being provided is intelligence, it would be great if it were 100x smarter, but it won't be. We will see the standard benchmarks go up a good amount but if it was truly 100x as intelligent as GPT-4x it would score 100% on every single test (which we all know it won't). Anyway, I feel like that chart is misleading simply because that 100x could refer to basically anything—training data, training cycles per unit of data, price reduction per query, server memory footprint reduction, who knows what—and it makes everyone imagine something 100x smarter than GPT-4x, which is pure fantasy. I like OpenAI but boy are they testing everyone's patience at this point.

[anta-zjbw]

i still dont even have the new voice engine yet

[BrianMosleyUK]

I wonder if Anthropic will maintain their lead with a Claude Opus release this year?

[WildBerryAura]

real talk, i am tired of them telling us fantastical things but showing us nothing

[k.k]

I dunno, they need to really update chat gpt, i find myself using claude far more than chat gpt.
Ill even use grok which gives slightly better answers for things than chat gpt, they need to update its context window and consistency, it drops information as youre building a project and get things flat out wrong, and even if you correct it, it still gets it wrong, but if you correct claude and guide claude to the issue, claude tends to understand and adjust.
Theyre really fixated on low cost and api support to be the first A.I to be implemented into devices and other small products, but if the a.I model isnt consistent it becomes a novel product.

[txzgmzp]

Gemini can't even give me the correct military time😂

[Lositosantos]

I am not mad that Ai reporters are making money off of speculation. Just wish yall could push out factual reports.

[macheteishrecords]

19:40 are there any dregs in the server? If there aren’t any that would likely change the sentiment of the rest of the community.

[BBD-JITZ]

Can’t wait for the video game applications

[Agoncor]

No GPT significant update will be done before the November 5 election in the US.

[Cory-vw]

This is what EQ wants. We want Non-locality & Observer-Dependent Reality. This relationship is going to go beyond human sentiment.
Moreover, quantum mechanics introduces the idea of entanglement, where particles become interconnected in such a a way that the state of one instantly influences the state of another, regardless of distance.
It is not my fault. Those humans' IQ use to be probabilistic, but not anymore.
See calcium-48 magnetic mystery & the pleasure will be all mine.

[tiagotiagot]

The "emergent democracy" actually just being a predefined president based on current real-life candidates seems very misleading and like engagement-baiting. I need to check to actual paper to see if it is just being misreported or of if they're themselves being disingenuous...