AI Burns

Basically think of the o3 results as validating Douglas Adams as the science fiction author most right about AI.

When given longer to think, the AI can generate answers to very hard questions, but the cost is very high, and you have to make sure you ask the right question first.

@dnystedt “Earlier this month, Broadcom announced it is collaborating with three major cloud service providers – likely Google, Meta, and ByteDance – on the development of AI chips, as noted by The Elec.”
https://www.trendforce.com/news/2024/12/20/news-sk-hynix-rumored-to-deliver-major-hbm-order-to-broadcom-in-2025/

Test-Time Compute scaling but in simple! @OpenAI o1/o3 made big waves by being able to scale inference compute relative to downstream performance.

Here is a poor man’s recipe for it. “Scaling Inference Compute with Repeated Sampling” is a paper that demonstrates how repeated…

Our co-founders discuss the past, present, and future of Anthropic.

Timestamps:
00:00 Why work on AI?
02:08 Scaling breakthroughs
10:57 Sentiment shifting
18:30 The Responsible Scaling Policy
30:42 Founding story
39:08 Racing to the top
43:43 Looking to the future

Interesting new research from @AIatMeta on exploring better reasoning using latent space instead of output tokens. “Coconut” (Chain of Continuous Thought) enables more efficient and effective reasoning, particularly for complex planning tasks.

In Coconut, the LLM has two modes.…

This is wild.

Hume AI just announced a new AI voice model.

It’s like ChatGPT Advanced Voice Mode, Elevenlabs Voice Design, and Google NotebookLM in one.

t can create a voice with a whole personality from a description or 5 second clip. And more.

8 wild examples:

Yesterday I shared you a prompts thread about waterfalls done with @freepik Mystic .

The quality is insane and it’s easier to animate them after ..

Workflow for the video :

Image : Freepik Mystic
Animation : Kling
Sound : Suno

Since the beta test and the launch creators…

Triad for AI video, August 2024
MidJourney for image
Runway for video
Udio for sound.
A fourth would be a surprise: ByteDance CapCut for editing, getting popular outside China
(Yes, other programs work fine)

Memory, not GPUs, looks to be the cost driver of AI systems. Extraordinary forecast & explainer by
@dylan522p, Jeff Koch, Tanj, @ChuWega
https://www.semianalysis.com/p/the-memory-wall Memory prices haven’t fallen as much

“The AI ​​model called ‘GPT Next’ that will be released in the future will evolve nearly 100 times based on past performance. Unlike traditional software, AI technology grows exponentially.” @tadaonagasaki Lead of OpenAI Japan. https://www.itmedia.co.jp/aiplus/articles/2409/03/news165.html @yleCun disagrees

Wait a minute my AI finishing songs I made in 2021!! Do you know how much unfinished music we all have as producers ?? the future is hear next level let’s talk about it

The first-ever autonomous AI agent civilizations have emerged — in Minecraft.

> 1,000+ agents building their own economy, culture, religion, and government in a virtual world

Imagine the potential of models like this at scale, simulating and testing solutions to real-world…

Take a look at the options for getting the face right