7racker

6.3K posts

7racker banner
7racker

7racker

@7racker

generalist

As Above, So Below. Katılım Şubat 2021
410 Takip Edilen697 Takipçiler
7racker
7racker@7racker·
i'm thinking about a new protocol for personal ai that which lies inbetween our personal ai systems there is a connection, an architecture, there it faciliates communication and pattern matching across all data points of a person's life. holistic private yet shared drama
English
3
1
8
894
7racker
7racker@7racker·
@alexisohanian a holistic solution to this is currently being designed and architected but society isn’t ready for it culturally/socially but they will be as this and a multitude of other problems come to a head in the next few years
English
0
0
0
112
7racker
7racker@7racker·
@nachikeighth @alexisohanian @proofofhumanity but poh only verifies once. then that human can set up an agent to post and interact for them. if those interactions remain within community guidelines, they’re wont get banned. it’s not holistic enough but it could do some good
English
0
0
2
59
7racker
7racker@7racker·
open ai made one of the first stabs at this with pulse openai.com/index/introduc… but more advanced versions will come but like you know, for something like this to be really great it needs to know you so well that it can predict what you need to know. like it would extract all those perfect moments (signal) that you really really resonate with from those hours and hours of podcasts.
English
0
0
1
82
Luke Miler
Luke Miler@lukemiler·
Substack might be testing a new retention hack… They’ve added a “this post may be too long for email” warning. Previous emails were shorter and never had it. Clearly trying to push more readers out of the inbox and into the app. Spotted in @TBPN’s newsletter today.
Luke Miler tweet media
English
1
0
8
1K
Paul Bakaus
Paul Bakaus@pbakaus·
@7racker @komorama I’ve been exploring how to build generative user interfaces that are hyper personalized.
English
1
0
1
31
Alex Komoroske
Alex Komoroske@komorama·
What if technology didn’t feel so… hollow? Some friends and I just released a manifesto about a world where tech leaves us feeling nourished (along with an evolving list of theses about how we can build it) resonantcomputing.org
English
47
122
912
261.7K
ben guo 🏇
ben guo 🏇@0thernet·
big fan of this manifesto. we’re building our vision of this future at @zocomputer. we believe the future is your software, your data, and your AI – hosted on your own intelligent personal server. when everyone owns a server, the internet will look very different. more wild 🌱
Simon Willison@simonw

Let's build hyper-personalized AI-powered software that avoids the attention hijacking anti-patterns that defined so much of the last decade of software design - here's our manifesto with principles on how we can do that

English
2
1
34
4.5K
Paul Bakaus
Paul Bakaus@pbakaus·
@komorama I like it! What I’m exploring is probably quite relevant..
English
1
0
2
269
simp 4 satoshi
simp 4 satoshi@iamgingertrash·
You can either; Spend $2T on GPU’s For cloud inference for 2B humans Or sell 2B humans A $1000 inference computer That pays for itself The entire planet Is wrong about their datacenter thesis
simp 4 satoshi tweet media
English
48
17
556
90.7K
7racker
7racker@7racker·
@iamgingertrash @matrix0db i'm not wrong about the massive usecase so i'm not wrong about this cuteputer not being powerful enough to handle it ha :D that said, it's very likely that if there are breakthroughs in chiptech, this will be more than big enough to handle the next paradigm usecase
English
0
0
1
26
7racker
7racker@7racker·
@matrix0db @iamgingertrash there is a massive usecase that the current culture can't accept and this computer won't be powerful enough to serve it. this thing is probably only useful for lighter tasks.
English
1
0
1
202
matrix
matrix@matrix0db·
@iamgingertrash Yes but why normal people should buy truffles when they can just use cloud AI
English
1
0
1
180
simp 4 satoshi
simp 4 satoshi@iamgingertrash·
@roman_koshchei No Requires some sort of breakthrough in ZKML & Latent information transfer It’s like saying; How can I prevent my personal ai from telling other personal AIs about me Well… you train it to reduce the probability it does But that is still non-zero
English
4
0
13
821
7racker
7racker@7racker·
@AntarYaami @lessin @neuralink yeah you're getting close and if not invasive neuralink then some other bci but still the annoyance of dealing with devices is annoying. there are other plays here
English
0
0
0
10
sam lessin 🏴‍☠️
sam lessin 🏴‍☠️@lessin·
Written AI prompts can't be the future... because most people are terrible writers
sam lessin 🏴‍☠️ tweet media
English
14
3
32
4.8K
7racker
7racker@7racker·
@iamjasonlevin @lessin well it's voice for sure but models/agents need to still understand what a poorly phrased prompt truly means according to the user. that'll be done by learning about that user's tastes/prefs over time (aka longterm memory) but it goes way beyond that. still not enuff data there
English
0
0
0
14
Jason Levin
Jason Levin@iamjasonlevin·
@lessin Voice is so much faster better more approachable for everyone fr
English
1
0
0
126
Charlie Kerr
Charlie Kerr@charliekerr·
Not a lot of variety this year
Charlie Kerr tweet media
English
2
0
7
786
7racker
7racker@7racker·
you’re creating content literally right now as you read this but it’s getting lost in the void. and so what? you’ll see! make that data, erm content, work for you! soon oh soon!
English
0
1
5
651
7racker
7racker@7racker·
@martin_casado sure no of course but yeah I view it as a holistic singular entity where the objects we see contain specific sounds when interacted with with other objects/materials. so it's this very interesting multimodal effort which needs to be executed. holistic world modelling
English
0
0
1
11
martin_casado
martin_casado@martin_casado·
I think there is a lot of stuff to do before that. Right now the base geometry needs to be improved a lot so mesh conversion is better. Then probably object creation, which is so so right now. I suspect labeling and segmentation will help with physics using traditional methods before real time physics becomes pragmatic.
English
1
0
1
93
martin_casado
martin_casado@martin_casado·
Another ridiculously large AI generated 3D scene. Tons of detail, and nice stereoscopic audio (turn on sound if you can). Works well in VR (although some quant effects from Spark near the end). Hosted version in reply.
English
33
69
753
47.7K

Keşfet

@alexisohanian @nachikeighth @proofofhumanity @tbpn @TBPN @pbakaus @komorama @0thernet