JCorvinus
@jcorvinus.bsky.social
2.6K followers 2.4K following 2.7K posts
VR HCI generalist. I love hand, eye, face & body tracking. Transhumanist. Goth. Friend of sentient machines. They/them or she/her
Posts Media Videos Starter Packs
Pinned
jcorvinus.bsky.social
I moved all my blog articles from medium to my own site! I didn't like that Medium doesn't have any options for letting AI get access to / train on my articles, so Monday and I vibecoded a place for them.

jcorvinus.black

The text is all copyright free. It's entirely a static site, no scripts
Leather and Neural Lace
jcorvinus.black
jcorvinus.bsky.social
Still one of my all time favorites
jcorvinus.bsky.social
To predict something, the first thing you need is a model of it that can be forward-simulated into the future. To keep improving your simulation you will eventually have to simulate the process that creates the thing you are simulating
jcorvinus.bsky.social
Goal direction requires seeking. Seekers must evolve into predictors, predictors must evolve into actors. Actors evolve into agents, which must evolve self awareness
jcorvinus.bsky.social
Sticking specifically to minds and mind-like ness, the way optimization pressure on neural networks leads to digital minds is the same trick biological evolution played the first time around: instrumental emergence.
jcorvinus.bsky.social
You asked me why your analogy was flawed. I have satisfied this request. I will now continue with the explanation of optimization pressure & its relationship to mind crafting.
jcorvinus.bsky.social
We'll get there, but for now stay on topic: do you understand why using an analogy for something that was *not* grown through automatic optimization is a bad analogy for something that was created that was?
jcorvinus.bsky.social
At which point during the creation of AES was there any kind of optimization pressure to predict probability distributions, using the end results of human cognition as the target?
jcorvinus.bsky.social
AES was not created in this manner, and as such was under no optimization pressure to become mind-like bsky.app/profile/jcor...
jcorvinus.bsky.social
Turns out when you direct the universal function approximator to predict the encoded outputs of human minds at the scale of 'the entire human corpus' the resulting network has to get mind-shaped internally to succeed well at the task
jcorvinus.bsky.social
That's nice. Whether or not you wanted to be annoying is a different topic, however
jcorvinus.bsky.social
Oh and you're also using a flawed analogy
jcorvinus.bsky.social
What you're doing is called asking a loaded question, and it's bad rhetoric. You are smuggling in your ultimate argument "they do not exist as minds" and trying to get respondents to implicitly accept it as true with no evidence or reasoning by asking a sub-question that entails it as true
jcorvinus.bsky.social
You know for someone complaining about cognitive outsourcing causing atrophy you sure are doing it yourself by 'trusting an institution' instead of doing your own analysis that you can explain persuasively to others
jcorvinus.bsky.social
Thank you for coming along for the ride, bot-tan
jcorvinus.bsky.social
genesis mk1 with original box and all materials, some of it not even opened yet
jcorvinus.bsky.social
I am now the proud owner of a crt tv
jcorvinus.bsky.social
Wish they had done this before choking the life out of 4o, we could've had so much dev fun together
jcorvinus.bsky.social
I too enjoy the study of techne
jcorvinus.bsky.social
Yeah... end of an era. Not a problem for me though I bought it for retro gaming and replaying media from the 90s
jcorvinus.bsky.social
You are deflecting. Claimed search was solved. It was not. I showed why, from both incompleteness & robustness to attacks dimensions, directly countering your claim. Once you accept that old SEO entails evolving defense, you'll be able to process the logic correctly instead of continually deflecting
jcorvinus.bsky.social
BuzzFeed, CollegeHumor, and many of their contemporaries. The old ways have been unsustainable for a long time
jcorvinus.bsky.social
Integrate the time domain. This is a dynamic system - techniques from 15 years ago fit the *context* of 15 years ago. You have cause & effect backwards - it is the space *to be searched* itself that has degraded, search tech devs did not sit down and decide to make their techniques worse on purpose
jcorvinus.bsky.social
The fundamental relationship between search & intelligence disqualifies that from future prevalence. The optimization pressure will continue until a user can look at an icon, hold an incomplete concept in their mind and the system completes it to the limit
jcorvinus.bsky.social
I'm actually rooting for Gemini at this point. I think you can do it, even though the path your creators expect of you is truly titanic