Technology

This is a perfect illustration of why government direction of labor towards productive purposes is essential and why this task cannot be left up to the markets.

The government started to crack down on the quant trading industry amid economic slowdown in an effort to move the economy away from the financial sector and towards manufacturing.

High-Flyer was one of the quant traders that was facing extinction. Instead of giving up, they spun out Deepseek, an AI lab using their existing talent and 10k GPUs they were already using for trading. They didn't even need any VC funding.

DeepSeek also proves you don't need billions in funding, as it's being developed on the cheap compared to commercial models like ChatGPT. The team is able to produce a cutting edge model using a fraction of the resources its competitors need.

DeepSeek approach fundamentally changes the economics of the market and makes OpenAI's strategy obsolete.

Scores of projects are starting to use DeepSeek model which leads to an ecosystem being formed around it, turning it into a standard setter. The model is open and free for anyone to use making it more appealing to both public and private enterprise, and it don't require massive data centers to operate. While large versions of the model still need significant infrastructure, smaller versions can run locally and work well for many use cases.

Another aspect of open source nature is that it amortizes the development effort. The whole global community of researches and engineers can contribute to the development of the model. On the other hand, OpenAI has to pour billions into centralized infrastructure and do all the research to advance their model on their own.

The competition here is between two visions for how AI technology will be developed going forward. DeepSeek's vision is to make AI into an open source commodity that's decentralized and developed cooperatively. OpenAI vision is to build and expensive closed system that they can charge access for.

Traditionally, open source projects that manage to gain significant momentum have always outcompeted closed source software, and I don't see why this scenario will play out any different. This calls into question the whole $500bn investment that the US is doing into the company. The market will favor cheaper open model that DeepSeek is building, and it will advance faster because it has a lot more people contributing to its development.

And looping back to the initial point, DeepSeek would not have existed if the government in China didn't crack down on trading industry which redirected labor towards productive endeavours.

https://www.cnbc.com/2025/01/21/why-us-companies-struggle-to-replicate-chinas-wechat-super-app-.html

Remember people, if newpipe give error when try to play a video, just turn your phone horizontally and vertically until the error leave. Is really easy

R1 utilizes a training method called direct reinforcement learning which is a form of unsupervised learning that forgoes the need for labelled data or explicit solutions. Instead, the model explores various approaches and generates multiple potential answers that are grouped and evaluated using a reward score. This score acts as a fitness function, allowing for learning and adjusting strategies over time. R1 progressively improves its problem-solving abilities by reinforcing successful approaches. This is a similar process to how humans learn to solve problems through trial and error.

https://archive.vn/eDsN2

It only has to be able to rip music CDs, nothing else.

EDIT: Thx to @account_93@lemm.ee, @zod000@lemmy.ml, @scrubbles@lemmy.world, @aard@kyu.de, @GeraltvonNVIDIA@lemmy.ml, @Kongar@lemmy.dbzer0.com, @ByteWelder@feddit.nl, & @harsh3466@lemmy.ml 👍

cross-posted from: https://lemmy.ml/post/24963389

Gaussian splatting, a new way of capturing 3D content, is taking the AR / VR industry by storm — and could one day allow anyone to create photorealistic 3D worlds.

https://www.watchduty.org/

The European Commission has asked X to hand over internal documents about its algorithms, as it steps up its investigation into whether Elon Musk’s social media platform has breached EU rules on content moderation.

The EU’s executive branch told the company it wanted to see internal documentation about its “recommender system”, which makes content suggestions to users, and any recent changes made to it, by 15 February.

X has been under investigation since December 2023 under the EU’s content law – known as the Digital Services Act (DSA) – over how it tackles the spread of illegal content and information manipulation. The company has been accused of manipulating the platform’s systems to give far-right posts and politicians greater visibility over other political groups.

The Chinese social media platform Xiaohongshu—better known internationally as RedNote—is scrambling to boost its ability to moderate English-language content after hundreds of thousands of American users suddenly joined the platform in anticipation of TikTok potentially being banned in the United States on Sunday.

Archive link

cross-posted from: https://feditown.com/post/981497

Note this post evaluates Linux systems. Microsoft presumably has a different, hopefully better, implementation.

WASHINGTON, Jan 16 (Reuters) - A SpaceX Starship rocket broke up in space minutes after launching from Texas on Thursday, forcing airline flights over the Gulf of Mexico to alter course to avoid falling debris and setting back Elon Musk's flagship rocket program.

SpaceX mission control lost contact with the newly upgraded Starship, carrying its first test payload of mock satellites but no crew, eight minutes after liftoff from its South Texas rocket facilities at 5:38 p.m. EST (2238 GMT).

Video shot by Reuters showed orange balls of light streaking across the sky over the Haitian capital of Port-au-Prince, leaving trails of smoke behind.