Death is walking through the ruins, whispering the crying souls: "Nobody is safe in the face of bloodthirst and revenge, I am merely their messenger, their slave."
@drumm_colin This makes it a very wide basket (which might not be wrong), but instinctively, for me, usually these norms are formulated around a system of worldviews connected to sacred/divine/supernatural ideas. Wouldn't it be a distinction worth keeping?
@drumm_colin I find it hard to wrap my head around the idea that being a religion is the only way and maybe I don't understand really how it operates. Why do you think so?
Someone at a coffee shop in central Israel noticed a religious Israeli wearing a yarmulka with the flags of Israel and Palestine and called the Police on him. The Police detained him, cut off the Palestine flag from the yarmulka and then released him. https://t.co/Z6qi72GLwD The Israeli Police under Ben Gvir has largely stopped upholding the law (actual crime is out of control) and is functioning as a right-wing repressive organ.
When the argument focuses solely on LLMs, it overlooks a crucial point: the vast amount of engineering within and surrounding these models blurs the boundary of where intelligence can be said to reside, if anywhere. This complicates both sides of the debate.
Once again, regardless of whether you think that ChatGPT understands anything or not, I think this argument is confused. To say that it can't possibly understand anything because it was only trained to "predict the next word" is just as idiotic as saying that humans can't understand anything because they were "trained" to survive and spread their genes.
This line of argument seems to boil down to the idea that, unless something works roughly in the same way as the human brain, it can't really be intelligent, but just as the same software can run on very different types of hardware there is no reason to think that human-like intelligence couldn't be implemented in very different ways.
Both sides tend to assume a clearer locus of intelligence than actually exists. In practice, an โLLMโ is rarely just a standalone model. It is embedded within a broader system: retrieval mechanisms, tools, memory layers, orchestration logic, safety filters, and interface design.
Anthropic said Mythos was too dangerous to release. Then four random guys in a Discord gained access on day one by guessing the URL...
This is pretty insane:
โ Group in a private Discord guessed the endpoint from Anthropic's naming conventions
โ They figured out the conventions from the leak in the Mercor breach three weeks ago
โ Used a contractor's legit eval credentials to walk in
โ Have been using it ever since to build simple websites
The AI that finds zero-days in every operating system on earth was defeated by address bar autocomplete... big yikes
@burkov People are putting an emphasis on the llms themselves, but in fact there is a lot of engineering work happening around these llms to make them even usable