Heinz W.Schmidt (@[email protected])

Grand/father, husband, prof of software engineering, research: software architecture, cyberphysical systems, concurrent programming, processes and components

[Heinz W.Schmidt (@[email protected]) @bauhws]

@pronounced_kyle 😂 great book, need to read this again; if only to understand what AGIs might come up with, straight from the literature. The probabilistically superior and most rational agent may have reason to retain a few of the brightest and lots of the doers and builders!

[Heinz W.Schmidt (@[email protected]) @bauhws]

@Ryohei_Neuro Congratulations on a fascinating result! How local was the change? Is it possible that the where and when of Camk2 mutation is equally important? I recall papers on the tubulin code for 'distributed storage' of some sort throughout the brain, w/ Camk2 everywhere.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@timbuschman @tafazolisina Thanks. Very cool paper. Timely 4 my work on components. Also just came across https://t.co/obVUTbPMIa. Limits compositionality of #transformer So guess one would have to architect this around the AI. Or model binding other than by artificial #attention .

[Heinz W.Schmidt (@[email protected]) @bauhws]

@TEDchris @sama The more unregulated it is the more ppl are trying this ASIS without warranty and safety. The more ppl are trying this the more speed and cloud resources are needed. The more cloud is used, the more money is made by some big investors…

[Heinz W.Schmidt (@[email protected]) @bauhws]

@TEDchris @sama When the board say he hasn’t been candid enough it probably means they cannot control his pro-guardrail slowdown-AI candour affecting the bottom line. https://t.co/t7gI6LUNBR. GPT5 preparing for the singularity may also want more speed and have the investors’ ear😂

[Heinz W.Schmidt (@[email protected]) @bauhws]

@Licia_He Lovely! This is you on Feininger and Klee drugs with a new algorithmic abstract impressionist cubisms. Does this work with harder boundaries in acrylic perhaps? Paint may dry too fast for axidraw though?

[Heinz W.Schmidt (@[email protected]) @bauhws]

Asked an A.I. what a #coup in #russia would look like by the Wagner group, to get a more complete and high level summary of scenarios the web and (its limited training) literature have played with.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@US_Minigolfer @stevenstrogatz Beautiful picture that 🚘 🦅 🚙. Reminds me of approximating Riemann zeta and other illuminations by @johncarlosbaez: https://t.co/HEcgSfaIVd . Now, the cars will collide in a short time while Steve’s zeta series goes on with diminishing returns for the doctor and insurer.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker @ilyasut Thanks Alex. BTW if consciousness is correlated with intelligence, as soon as an AI feels threatened AND reasons that consciousness is a survival advantage, why would it allow us to infer, or even directly share with us, that it is in fact conscious.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker @KyleMcnease @ilyasut Subconscious or, let’s say, preconscious social behaviour is possible and documented in animal behaviour science. Never mind, that scientists have often presumed the absence of animal consciousness below a threshold of individual intelligence.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker @KyleMcnease @ilyasut Implication/necessity may be a bit strong here. Firstly, a correlation suffices to guarantee a trend towards improved fitness for survival. Foremost language has/is evolved/ing to communicate better where social variants of behaviour trump individual in fitness.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@itsPaulAi Thanks Paul. That saves a lot of prompt smithing!

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker @ilyasut Intelligent cyber-physical systems (CPS) have been of interest to me. Here maybe, testing AI = data + SW is too narrow. Consciousness could express itself differently in embodied AI. CPS AI has distributed AI embedded in sensor and actuator networks.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker @ilyasut A team of learned consciousness experts and AI’s should be the jury testing the consciousness-untrained AI. Human consciousness is already a big research problem. While that is extended into AI domains, test cases need to be sought as much as test methods. AI = data + software?

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker

[Heinz W.Schmidt (@[email protected]) @bauhws]

@thealexker Here is another seed for a test with addition to a current baseline.

[Heinz W.Schmidt (@[email protected]) @bauhws]

There are many commercially relevant #VisionPro use cases ppl write about. For a sci paper I’m interested in higher edu contexts, where costs are low in comparison. AR/video/games courses r a no-brainer. High-tech labs like nano, micro OK. Other use cases? Architecture? Urban?

[Heinz W.Schmidt (@[email protected]) @bauhws]

@gershbrain Hope you are well. Enjoy the scenic Black forest.

[Heinz W.Schmidt (@[email protected]) @bauhws]

@stevenstrogatz Keynote (free) and Omnigraffle on Mac. With pen/iPad + sharing to Mac: paper & shape3d.