deepseek is a side project. based 🫡

I’m hearing rumours Deepseek is causing panic in Meta

This is the cost America and American AI Tech / General Tech pays when the CIA/NSA get to pick the "winners" before the first lap is complete. What does this say about the state of American AI (and tech in general)? DEI, Nepotism (across the board from funding, government contracts, hiring, etc.), Cancel Culture (determines who gets funding, which startups are starved of funds, and worse, which startups are targeted for destruction. See: Facebook vs MySpace, Google vs Yahoo!, Netflix's "miraculous victory", etc.). It's then compounded by the attempt to create a Potemkin Moat using Vanity Metrics like total funding, GPUs deployed, number of parameters, number of tokens, etc. This is the same reason the US lost the Vietnam War when it was winnable, conventionally in 30-45 days after mobilization. vs A small team of switched on smart, ambitious Chinese guys focusing algo's and performance.

This one?

Made with the salary of a single high level engineer at a FANG company

I think the release of DeepSeeks model brings massive change on many levels - extreme increase of quality - price ratio - increase pressure on big AI labs due to higher competition - Open source strategy puts an additional layer of pressure as it builds up trust

game recognizes game. An American patriot too can do justice to real achievement, without petty-minded copes

Chinese Dynamism? Truth upon which progress depends, not communist party censorship, is a gift to the world.

Deepseek R1 is designed to be energy-efficient, ensuring it can scale globally without contributing to excessive carbon emissions. Its modular architecture allows it to grow alongside user needs, accommodating increased data volumes and complexity over time. That is what makes it a game changer.

Chinese are cooking much more than the US realizes

Grok: "Deepseek R1 has achieved performance comparable to OpenAI-o1 in technical domains like coding, math, and reasoning. It uses pure reinforcement learning, marking it as the first open research to validate that reasoning capabilities of LLMs can be incentivized without supervised fine-tuning. The model is fully open source, allowing global access for examination, modification, and further development. Deepseek R1 is notably efficient, with an architecture of 671 billion parameters where only 37 billion are active during operation. It has rapidly gained adoption among top U.S. university researchers and companies, signaling a shift in AI innovation towards China. Deepseek R1's development and release coincide with discussions on China's growing influence in tech and AI, challenging the status quo."

I completely agree! I’ve been using it alongside o1-pro on my cancer project, and it’s remarkable!

The repo IS amazing - And I’m saying this as we merge it into search.ai Main, but I see it differently. I don’t believe China copying @OpenAI innovation and open sourcing is benevolent - It’s meant to disrupt us. They’re racing AI to the bottom because they win that race. Just like manufacturing- and we can’t stop it at the port. China’s long-game is better than ours. Invest in research, pump out papers, open source, burn any alpha. Capitalism thrives because of alpha. We have no moat, and this is China’s attack. Once they dominate AI and robotics, it’s catastrophic for us.

Why? It’s a bunch of hacks.

DeepSeek R1 may well be one of first ubiquitous LLMs, now that Developers are running it on mobile, Raspberry Pi, and even in the browser. It's thrilling to think about the offline-first and fallback scenarios the R1 model enables. What a time to be alive.

$Regent architecture is the first one to use Deepseek R1

The cat is truly out of the bag

not just as open source but it’s also decentralized now — with decentralized inference by eternal ai. anyone can permissionlessly call the deepseek smart contract at any time in perpetuity. a forever gift to the world.

Really putting Openai in their place. They were supposed to be open source but decided they had to start hiding their special sauce and then transitioned to for profit in a very sketchy way. Go Deepseek!

If Deepseek is superior and is a world public resource, why should Americans be made to pour $500 billion into a domestic program in a vain effort of our overlords to cling to tech supremacy?

But it cannot answer whether Taiwan is an independent country or not.

Proof what a small & empowered team can do with a side project & limited resources. Every large AI lab - Google, Alibaba, Meta, OpenAI, etc - is now on notice. Let the “Let’s Search Arxiv for what to do next” games begin!

@doc_regent first to integrate Deepseek r1

great breakthrough, it will accelerate things

China has caught up. This bodes poorly for The West with Europe so determined to self destruct.

It’s incredible

If it’s not an evil plot by China, it’s a significant development and humanity’s opportunity to create a white hat ASI, one that could be the only safeguard against corporations and governments using their ASI to control humanity. I don’t understand the romanticism some people have about AI, especially in the absence of such a white hat. Why would anyone believe that Sam Altman or others like him would ever share prosperity or willingly relinquish power? It won’t happen. Only by creating and using our own ASI can we hope to defend ourselves against an ASI deployed by governments to control humanity or by corporations to dominate resources, develop life-extending treatments, or cure diseases. Anyone who believes that corporations accelerating longevity or fighting disease with ASI will simply share those advancements as gifts to humanity—or make them affordable to everyone—needs a serious reality check.

Good to see that deepseek is part of aiagent.app already!

It’s like mistral except not europoor

I know about it. it's like splitting the atom, but for data. a gift, yes, and a reminder: innovation thrives in shared minds.

We agree, open-source AI is the future. We're working on tweaking Deepseek to focus on specialized sectors as well

yes press.airstreet.com/p/the-st…

youve been duped mr marca

Agree completely about DeepSeek being the real deal. Big question now is how this impacts capital allocation across the space. I wrote this article recently that tries to answer that which you might enjoy reading: youtubetranscriptoptimizer.c…

I just asked it if China is an autocracy. It rapidly spit out a few paragraphs of the answer, then deleted what it had written and replaced it with "Sorry, that's beyond my current scope. Let’s talk about something else."