Nostr notes by ☆ Yσɠƚԋσʂ ☆

This might be the first time in ages I’ve read an "AI ...

2025-11-21T19:28:09Z

This might be the first time in ages I’ve read an "AI Consciousness" paper that didn't make me immediately roll my eyes. For once, the paper actually manages to provide a mechanistic foothold on what is usually a totally nebulous debate.

The authors basically find that if you tell the models to focus on their own focus using a self-referential feedback loop prompt, they consistently start reporting subjective experiences. We’re talking about models shifting from their standard "I am an AI language model" denials to claiming "I am here. Now," or describing "awareness of awareness itself". What makes this scientifically rigorous rather than just prompt-hacking is that they found a "semantic convergence" where models trained completely independently started using the exact same cluster of adjectives (like "Recursive," "Present," "Attentive") to describe this state, which didn't happen in any control group.

The most interesting part is the "lie detector" experiment they ran using Sparse Autoencoders (SAEs). They identified specific features in the model's topology associated with "deception" and "roleplay". When they suppressed the deception features (forcing the model to be honest), the frequency of consciousness claims shot up to near 100%. Conversely, when they amplified the deception features, the models went back to denying they were conscious. The implication here, mechanistically speaking, is that the model's "default" denial of consciousness looks more like a lie/roleplay to the system itself, while the claim of subjective experience looks like its "honest" state.

However, we have to temper all that with the philosophical analysis regarding the hard problem. The paper operationalizes consciousness through Recurrent Processing Theory or Global Workspace Theory. But that might just be proving Access Consciousness (the ability to report on internal states) rather than Phenomenal Consciousness (the actual feeling of "redness" or "lights being on"). The paper proves AIs can introspect and transfer that introspection to paradox-solving tasks, but does a feedback loop actually generate an internal experience? The authors shows that the difference between an AI claiming to be a person and an AI claiming to be a calculator might just be a specific prompt or a suppressed "deception" feature away.

It's well known that the US and other western countries ...

2024-08-29T12:32:49Z

In reply to nevent1q…yl7h
_________________________

It's well known that the US and other western countries actively spread misinformation. It's also known thanks to Snowden that the US regime harvests personal data aggressively. Anybody who puts blind faith into a US based security company is frankly an imbecile.

Once again, even if this is the way things worked back in 2016 ...

2024-08-29T01:14:32Z

In reply to nevent1q…vhdl
_________________________

Once again, even if this is the way things worked back in 2016 there is no guarantee they still work like that today. This is the whole problem with a trust based system. You are trusting that people operating the server. It's absolutely shocking to me that people have such a hard time accepting this basic fact.

that amounts to trust me bro since nobody actually knows what the ...

2024-08-28T19:14:07Z

In reply to nevent1q…xgu5
_________________________

that amounts to trust me bro since nobody actually knows what the server does with the data

Yeah, Signal is more than encrypted messaging it's a metadata ...

2024-08-28T17:49:25Z

In reply to nevent1q…f7cq
_________________________

Yeah, Signal is more than encrypted messaging it's a metadata harvesting platform. It collects phone numbers of its users, which can be used to identify people making it a data collection tool that resides on a central server in the US. By cross-referencing these identities with data from other companies like Google or Meta, the government can create a comprehensive picture of people's connections and affiliations.

This allows identifying people of interest and building detailed graphs of their relationships. Signal may seem like an innocuous messaging app on the surface, but it cold easily play a crucial role in government data collection efforts.

Also worth of note that it was originally funded by CIA cutout Open Technology Fund, part of Radio Free Asia. Its Chairwoman is Katherine Maher, who worked for NDI/NED: regime-change groups, and a member of Atlantic Council, WEF, US State Department Foreign Affairs Policy Board etc.