saved
astonishing blogpost from Shengyu Liu
Hraness cites a source capture. The source author remains the source.
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
astonishing blogpost from Shengyu Liu (刘胜与, also known as interestingLSY/intlsy), kernel engineer at DeepSeek. The first part is his personal struggle with the fact that his work is about to render his beloved craft obsolete. The second is… well. let's just say we agree.
Perhaps these are questions that only history itself can answer.
Conclusion
As AI develops, the society of the future may drift toward one of two extremes: communism or Cyberpunk 2077.
In the former, productive forces are liberated to an extraordinary degree, and people’s living standards improve significantly. (I’ll stop there, or I’m afraid this might not make it past the censors.)
In the latter, a small number of technology companies control most of the world’s resources. Only a tiny minority of people can access the most advanced AI and other technologies, achieving something close to “mechanical ascension,” while the majority are left with only weak and crippled AI systems.
Moving between social classes will become increasingly difficult: you would need access to the strongest AI in order to rise into a higher class, creating a vicious cycle.
Suppose Anthropic permanently controls the most advanced AI in the world.
Do you think the society of the future will become communism—or 2077?
Take a guess.
Go on.
That is why I still believe that frontier intelligence should be made available to everyone in an open and inexpensive form.
I do not trust Anthropic or OpenAI to do this. In particular, I do not want Anthropic to control the world’s most advanced artificial intelligence or AGI. To put it dramatically, I consider the stakes comparable to Hitler obtaining atomic-bomb technology before the Allies.
That is also why I chose to stay at DeepSeek, and why I continue to stay.
We research powerful, fast, broadly accessible artificial intelligence and release it as open source. Perhaps, in doing so, we can pull the world a little farther away from the 2077 end of the spectrum.
May all be well in the world to come.
May all the beauty be blessed.
[1] “Main Attention” here refers only to MQA attention with head dim = 512. It does not include the indexer used to select the top-k important tokens. That part was written by other colleagues—who are also extremely skilled—together with their AI agents.
