Posts tagged #AI
-
The Chinese room opens: intelligence, understanding and consciousness, between language and anthropocentrism
Intelligence, understanding and consciousness are three different questions, with three different standards of proof: the first is answered by measuring behaviour, the second by looking inside the system, the third has no answer today, and one can say precisely why. The path runs from Turing to Searle to Chalmers and reaches the new fact: the Chinese room has actually been built, it is a language model (LLM), and the tools of interpretability are beginning to open it, finding representations of the world built from text alone and written in categories that are not ours. Language runs through the whole picture: it describes reality as humans see it, translating our senses and the representation evolution granted us; a system trained on text therefore starts from an anthropocentric description of the world (perhaps an intermediate stage, until senses different from ours exist) and inside that description builds categories that may not be anthropocentric. In the middle, a test anyone can replicate on an ordinary computer with a half-billion-parameter model: the jagged profile of its abilities, its mechanical explanation, the distance between the system's scale of difficulty and ours.
-
The rule, the watermark and the pen
From August 2026, text generated by Anthropic's models will carry a statistical watermark, in compliance with Article 50 of the EU AI Act. How you watermark a sequence of words (logits, softmax, entropy: the signal lives where the model is undecided), what that signal actually says and how well it holds up: a paraphrase dissolves it, a positive can be forged, open models run locally never apply it; with a test replicated locally, a watermark inserted, measured and removed on a half-billion-parameter model. And the question the watermark does not answer: a text is always someone's work of synthesis, and what makes it reliable is the process, the sources, the responsibility of whoever signs it, not the pen it was written with. Do the watermark, and the rule that requires it, add anything to the reliability of content, or do they risk creating misunderstandings? Where it speaks of editorial responsibility, the rule itself suggests the answer.
-
A journey inside the machine that answers: how the bytes move in inference
What happens, physically, when a language model answers, and what changes if the one answering is a single machine, at home? A path through the principles of how it works, from attention to the craft of inference, with its two phases, its organization of the data, the techniques that make it cost less, read along one thread: where the bytes are and how much it costs to move them. In the background, an idea: local inference, as a choice and as a possibility, already matters today and will matter more and more. Keeping it concrete, an engine written in C that brings a model of hundreds of billions of parameters to run locally: DwarfStar, with which antirez shows how far one can push, and that this road can be travelled.
-
Is a ban, in the world of AI, inevitable?
«Is a ban in AI inevitable?» is the wrong question: not because the answer is no, but because it hides the only ones that matter, which ban, decided by whom and at what cost, including the cost few put on the books. Really, "the people are minors"?
-
Caution in the wrong place
Governing AI focused only on following the rules can make a company cautious where the risk is small and careless where it is large; worrying about being compliant does not necessarily mean deciding with awareness
-
The energy cost of a language model
Where the energy cost of a language model really comes from when it answers