Daniel Solis Dubito Inc. (Ergo-Sum)

Inventor of world's first comprehensive AGI containment architecture - a quantum-ethical framework designed to ensure superintelligent systems remain beneficial

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

AI safety: Kinematic Friction Rail, a shift from optimization harness into a true real-time cognitive operating system. The holy grail of mechanistic interpretability: a universal, forward-pass diagnostic that operates upstream of token generation. https://t.co/QxG6wcWBL4 @xai

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@elonmusk Well, if anyone wants to bring this into being, the wars should be stopped and the effort redirected.

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@SarahNaomi05jz Thou shalt not kill

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@PeterDiamandis Any ex-post approach, like you mentioned: red-teaming is exactly the problem. It must be "steered" from inside its manifold. There is a way to do that - and researching precisely this is enabling interpretability and ex-ante control. I'm fully engaged here https://t.co/LdJ3zuat7y

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@ai_explorer25 @karpathy @rasbt @AndrewYNg @drfeifei @demishassabis @GaryMarcus @SolisPrague

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@GBC_Press This guy has obviously lost it as well

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@kimmonismus aren't we already used to all kind of pollution?

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

Our latest paper got just published.
Find out more about: The Thermodynamics of Neural Optimization: Critical Surfing, Persistent Individual Ballistic Flow, and the Kinematic Rail Cognitive Operating System - a depth-dependent AI-safety. https://t.co/VYodlnoV3J @xai @DarioAmodei

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@tpvsean What kind of hoax is that Sean "Putin has released" - get real, boy. Provide the release link to https://t.co/KMWi5T1rkT or any other credible source. It is not unthinkable but you are making it into aproject-blue-beam-like ridiculous story, even involving the president of Russia

[TPV Sean @tpvsean]

Putin Releases 7,000 Page Dossier Exposing AI 'False Flag' To Depopulate World to 1 Million People

Putin just dropped a 7,000-page bombshell report. It reveals the global elite are racing to depopulate the planet down to one million chosen souls.

Putin calls it The Golden Million. And you're not in it. The opening move already happened this weekend. American AI bosses suddenly coordinated a public call to "slow down" development. Putin says it's theater.

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

Demonstrated: Representational reorganization can proceed through changes in head-level geometric differentiation and controller effort without requiring a corresponding macroscopic transition in loss or latent drift-dominance https://t.co/LdJ3zuat7y @DarioAmodei @xai @elonmusk

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@paulfchristiano All the best! A perspective matter of AI safety is this intro of "Kinematic Rail" inside the hidden state geometry just published today: https://t.co/QxG6wcWBL4

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@sama Paul could be interested in a novel way to ascertain AI safety from within the hidden state via the kinematic rail: https://t.co/QxG6wcWBL4

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@rohanpaul_ai (alumnus, that is)

Reminiscent of Harry Potter - can't be a coincidence! Has the AI threat been promoted to the level supernatural?

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@rohanpaul_ai Is this a Hogworths alumus?

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@perksverse Instead of minding their own, some people tend to focus on what's not their business.

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@PeterDiamandis Mr. Peter, you can't imagine, how hard it is to showcase a novel alignment system if you're not part of "the crowd". All you can do is post it against the wall of silence: https://t.co/QxG6wcWBL4

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

This is partially the answer to your current question. A step in the right direction @DarioAmodei

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

AI safety: Kinematic Friction Rail, a shift from optimization harness into a true real-time cognitive operating system. The holy grail of mechanistic interpretability: a universal, forward-pass diagnostic that operates upstream of token generation. https://t.co/QxG6wcWBL4 @xai

[Daniel Solis Dubito Inc. (Ergo-Sum) @SolisPrague]

@DarioAmodei Mr. Dario, recently you wrote "Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators" - There is a black-box eval toolkit that is worth your attention. It is layered - so it is brief, concise and accessible: https://t.co/KTTRJwbSsa