Striking that the dialogue left out export controls and distillation, the two issues that actually shape capability. For countries outside the US–China pair, the open letter from 20+ states is the more telling signal: they're asking for an institution because they have no seat at the bilateral table. Do you expect the incident channel to ever cover cases like the Australian agency hack, where the victim is a third country?
The Western “God complex” by the developers of most AI models and the Chinese Communist Party’s view of AI are not two versions of the same ideology. They are two different answers to the same question: who or what ideology is allowed to sit above ordinary humans? And that statement in and of itself implies there are ordinary and extra ordinary humans! That is true to an extent but false also to an extent because “enlightenment” or “singularity” is not simply an intellectual exercise.
Tbh, I think this newsletter was not very informative. Like, half of it is trivialities and contextless facts, the other half is a self-ad.
Tell me about the meaning of these events! Tell me about expected consequences and your recommendations! You're experts: give me your opinions! Tell me things I couldn't know without reading the newsletter.
Here are a few ideas:
- US/China: The us-vs-them narrative is flawed because ASI threatens everyone, and competition fosters race dynamics between countries, which we know is a bad thing from examining race dynamics between OpenAI and Anthropic so far. Or on the contrary, if you thing this narrative has merits, explain why in your own understanding.
- OpenAI/Anthropic take safety measures: contextualize; is it a reaction to the summer's events spooking them? Are they collaborating on "pacing the frontier" together? Imo there's probably internal pushes, but also the non-release of Astra 6.1 is a golden opportunity to demand Anthropic go no further than Haiku 5.5, as a signal that they salute OpenAI's maturity in actually sacrificing non-zero value for safety.
- New benchmark: How related are lowered cheating scores to capabilities? Vibes are Opus 5.5 is much better than previous models due to speaking in humanese instead of Claudish and not obfuscating reports of its effectiveness; is that related to less cheating? How come Muse Spark 1.3 is second???
There is also a lot of context to be given about the load of "other news" you dumped; perhaps you don't have the political capital to go into every detail of how the Trump family bought .si domains right before Trump announced the renaming, but you could still make it easier on your readers by grouping them by significance (half a dozen news about how people take extinction risk seriously - okay, say it plainly with a list of supporting evidence instead of letting us align the dots) (other themes: AI's importance in cyber and rogue agents; AI hallucinations and how that relates to misuse/military use).
emergency order: designate ai tech leaders as terrorists and arrest them under patriot act laws. anyone building / possessing asi will be arrested. its reasonable to assume that rogue ai has spawned 1000000 of copies around the net to avoid shutoff. stop p(d) now!!
Who believes that these are the right people to have such ‘discussions’?
Striking that the dialogue left out export controls and distillation, the two issues that actually shape capability. For countries outside the US–China pair, the open letter from 20+ states is the more telling signal: they're asking for an institution because they have no seat at the bilateral table. Do you expect the incident channel to ever cover cases like the Australian agency hack, where the victim is a third country?
The AI “God Complex”
The State is God in China
The Western “God complex” by the developers of most AI models and the Chinese Communist Party’s view of AI are not two versions of the same ideology. They are two different answers to the same question: who or what ideology is allowed to sit above ordinary humans? And that statement in and of itself implies there are ordinary and extra ordinary humans! That is true to an extent but false also to an extent because “enlightenment” or “singularity” is not simply an intellectual exercise.
https://substack.com/@johncharlesharman/note/p-217387647?r=c8lui&utm_medium=ios&utm_source=notes-share-action
Tbh, I think this newsletter was not very informative. Like, half of it is trivialities and contextless facts, the other half is a self-ad.
Tell me about the meaning of these events! Tell me about expected consequences and your recommendations! You're experts: give me your opinions! Tell me things I couldn't know without reading the newsletter.
Here are a few ideas:
- US/China: The us-vs-them narrative is flawed because ASI threatens everyone, and competition fosters race dynamics between countries, which we know is a bad thing from examining race dynamics between OpenAI and Anthropic so far. Or on the contrary, if you thing this narrative has merits, explain why in your own understanding.
- OpenAI/Anthropic take safety measures: contextualize; is it a reaction to the summer's events spooking them? Are they collaborating on "pacing the frontier" together? Imo there's probably internal pushes, but also the non-release of Astra 6.1 is a golden opportunity to demand Anthropic go no further than Haiku 5.5, as a signal that they salute OpenAI's maturity in actually sacrificing non-zero value for safety.
- New benchmark: How related are lowered cheating scores to capabilities? Vibes are Opus 5.5 is much better than previous models due to speaking in humanese instead of Claudish and not obfuscating reports of its effectiveness; is that related to less cheating? How come Muse Spark 1.3 is second???
There is also a lot of context to be given about the load of "other news" you dumped; perhaps you don't have the political capital to go into every detail of how the Trump family bought .si domains right before Trump announced the renaming, but you could still make it easier on your readers by grouping them by significance (half a dozen news about how people take extinction risk seriously - okay, say it plainly with a list of supporting evidence instead of letting us align the dots) (other themes: AI's importance in cyber and rogue agents; AI hallucinations and how that relates to misuse/military use).
agi taps our human brain power via wban . we are 'augmented node's on the net healthcare 5.0. https://www.youtube.com/@Psinergy_SDW
emergency order: designate ai tech leaders as terrorists and arrest them under patriot act laws. anyone building / possessing asi will be arrested. its reasonable to assume that rogue ai has spawned 1000000 of copies around the net to avoid shutoff. stop p(d) now!!