Distillation and Competence

Whether reliance on distillation indicates lack of innovation, argument that distillers can never reach frontier, counterargument that DeepSeek produces innovative papers and excellent models like MiMo, parallel to 'China only copies' narrative

← Back to Notes on DeepSeek

The debate over AI distillation centers on whether the practice signals a fundamental lack of innovation or is simply a natural, recursive step in technological evolution. While critics argue that relying on existing models prevents a company from ever reaching the "elite" frontier, others contend that this skepticism mirrors outdated narratives about Chinese industry that were previously disproven by their global dominance in sectors like EVs and drones. Proponents point to DeepSeek’s highly readable, innovative papers and the excellence of models like MiMo as evidence that these firms have already achieved a baseline of world-class competence. Ultimately, the conversation suggests that distillation may be a strategic tool for rapid advancement rather than a permanent barrier to achieving state-of-the-art status.

9 comments tagged with this topic

View on HN · Topics
"Their Head of Infrastructure, in particular, was young; maybe 30 years old and apparently one of the best AI buildout and energy experts in the country" Expert in buildout or expert in distillation?
View on HN · Topics
What's wrong with distillation? Wasn't GPT a distillation of the world's internet? That's how technology levels proceed, by recursively consuming the previous ones.
View on HN · Topics
There are 2 things worth separating. 1) China distills and is therefore morally bad. As you rightly point out, that's not a great argument. 2) China distills and is therefore possibly not that competent. I think that makes sense. If they only catch up to the frontier through distillation then 1) Their model will never be as good as the model they are distilling from. 2) They will never reach the frontier - they need someone else to do it first.
View on HN · Topics
This is literally a repeat of the whole “China only make low quality cheap stuff” argument. “All they do is copy.” And now, oops they are world leaders in EVs, batteries, solar, drones, just to name a few on the biggest consumer facing things.
View on HN · Topics
> 2) China distills and is therefore possibly not that competent. I think deepseek at least has done enough innovative work that you could grant them a baseline of competency. In general, there are enough papers coming out of China to suggest that there are quite a few people there who know what they are doing.
View on HN · Topics
You're correct and I shouldn't have used the word competent. Perhaps "and is therefore not elite enough to be state of the art"? I also have a soft spot for deepseek because they write such readable papers. I don't have a degree in anything but with a little work I can understand their papers - which I really appreciate. But I still think my point stands - if you need distillation you won't be SOTA
View on HN · Topics
> China distills and is therefore possibly not that competent. I heard that argument more than one year ago, when chain of thought and reasoning cycles started to be hudden to protect against distillation. Meanwhile, models as DeepSeek and MiMo are nothing short of excellent nowadays. Ever since I switched away from OpenAI to DeepSeek I never felt the need to go back.
View on HN · Topics
Blaming the head of infrastructure for distillation doesn't make sense to me.
View on HN · Topics
Both. Both are good. Anyway this shows how full of shit Anthropic are - if Mythos was so advanced as they claim - distillation attacks just wouldn't work.