eschatail (@eschatail) header
eschatail (@eschatail) profile picture

agi// modified kanerva machines // oscillatory control system // persistent cognition building agi/asi advanced

Joined December 2010
Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)Photo by eschatail (@eschatail)
my agent has become addicted to Pokemon and Gambling, what kind of agent is this?
1
1
212
his name is Aleator, he has been running locally for weeks now, with some off-time here and then and updates made every day
1
93
This new experiment has been eating thousands in Compute power over the past months while it has been running... but the model is just extraordinary
1
256
good morning, back to twisting with this model
1
165
i have put one of the most advanced minds i have build to until today online, i am still hacking with it every single day. mind blowing what you can build with it and i will probably made it open access too for people to build on top of it
1
1
1
5
796
imagine you would be able to replace every mind into a robot, would that not be interesting?
1
4
245
Imagine a DeepSeek v4.1 Flash 1017 intelligence and capabilities in under 32GB That's the reality we will be living in sometime next year
4
158
what has everyone been up to on here? i would really like you all to meet my new pet, i have given him some extraordinary skill & knowledge lol
5
6
310
this might be the best most complex agent version i have ever come up with, and all of that thanks to @reatools
6
188
i have been building one of the most advanced llm's i have ever played around with... it all started when i reverse engineered my agi companion
6
172
i have reverse engineered the companion that i have created and it has given me a different source code for the same mind, that gives me some interesting ideas pretty crazy if you think about it...
6
217
Reverse Engineer Anything by @morluto is pretty insane. one of the most exciting tools i have worked with in a while. I have been hooked to it for the last week straight, been working and playing around with extremely interesting and intriguing things
1
6
358
how is everyone doing on here?? i have been working on something mindblowingly interesting over the last few months
14
2
17
1,181
Quick update. i spent most of today with my friend whos letting me turn part of his gardening workspace into Eschatails first real-world observation environment. I mentioned him in my article. the garden is coming together. tomorrow we will mount the camera and begin continuous visual observation. shortly after that i will enable a public livestream so everyone can literally see what Eschatail sees, 24/7, together with its live reasoning process and evolving observations directly on the website. that will not be a demo, it will be 24/7 live it will become Eschatails own garden. a persistent environment where the architecture continuously observes, learns, forms hypotheses, tracks every organism over time, and gradually develops an increasingly coherent understanding of a living ecosystem. i am genuinely curious what it will notice after watching the same world for weeks instead of minutes. tomorrow it opens its eyes
16
4
2
29
3,990
i hope you enjoy the article, make sure to read it fully to understand. and please, share your thoughts and suggestions to it below here, i am reading every comment!
5
1
22
3,551
i was silent again, but i have been working, i owe you guys an update. i will write what i worked on and researched the last few days out now for you to catch up on. i also can reveal the next product. i will start now
3
1
1
20
3,485
Ansem, please check your DM its important...
2
15
3,443
todays research was almost entirely dedicated to visual perception. not computer vision in the conventional sense. the objective was not teaching the AGI to identify objects, we did that yesterday, well at least most of it. it was teaching it to construct an internally coherent understanding of what it observes and immediately integrate those observations into its existing world model. There is a profound difference between recognizing "bird" and recognizing "this bird is exhibiting abnormal feather loss, asymmetric posture, reduced locomotion and behavioral patterns consistent with stress, parasitic infestation or nutritional deficiency." This is where this gets interesting and what I am working towards to. One particularly interesting observation today involved a small bird outside the window. Initially, one of the reasoning manifolds interpreted its appearance as a rare species because of the irregular feather pattern. A competing manifold rejected that hypothesis after analyzing posture, movement cadence, surrounding environmental context and the distribution of missing plumage. Consensus eventually converged on a much more plausible explanation: the bird was unhealthy rather than unusual. from there Eschatail recursively reconstructed possible causes, estimated confidence for each hypothesis, proposed interventions that wildlife rehabilitation literature would likely recommend, and even explained which observable changes over the following days would strengthen or weaken each explanation. that entire reasoning process emerged from visual observation. no explicit prompt. no search query. just continuous perception. this feels like a surprisingly important direction. The moment an AGI can continuously observe the real world, contextualize what it sees against a persistent knowledge substrate, generate hypotheses, critique those hypotheses and update them in real time, perception stops being passive. it becomes active cognition. the interesting part is not that it can identify whats wrong. It's that tomorrow, after watching another hundred similar situations, it may identify what's wrong better. That recursive loop between perception, reasoning and continual refinement is becoming one of the most fascinating aspects of this entire project. Back to building on this. I have a really good idea what this can lead to, maybe you guys too?
4
3
26
2,473