--- Log opened Sun May 05 00:00:46 2024
02:31 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
02:31 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
02:57 < hprmbridge> Eli> They already use AI in the Ukraine war. Due to jammers, the drones use AI to recognize and kill the target during the final attack phase. It’s fairly effective apparently. I was shocked that they are apparently using over 10000 drones a month just on the Ukraine side.
03:01 < hprmbridge> Eli> https://www.cbc.ca/amp/1.7130368
03:13 -!- darsie [~darsie@84-112-12-36.cable.dynamic.surfer.at] has joined #hplusroadmap
03:23 -!- darsie [~darsie@84-112-12-36.cable.dynamic.surfer.at] has quit [Remote host closed the connection]
03:23 -!- darsie [~darsie@84-112-12-36.cable.dynamic.surfer.at] has joined #hplusroadmap
03:59 -!- stipa [~stipa@user/stipa] has quit [Quit: WeeChat 4.1.2]
05:04 -!- CryptoDavid [uid14990@uxbridge.irccloud.com] has joined #hplusroadmap
06:48 -!- geneh2 [~cam@pool-173-66-190-123.washdc.fios.verizon.net] has joined #hplusroadmap
08:10 -!- stipa [~stipa@user/stipa] has joined #hplusroadmap
09:05 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
09:05 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
09:16 -!- CryptoDavid [uid14990@uxbridge.irccloud.com] has quit [Quit: Connection closed for inactivity]
09:23 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has left #hplusroadmap []
10:37 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap
11:49 -!- justanotheruser [~justanoth@gateway/tor-sasl/justanotheruser] has quit [Ping timeout: 260 seconds]
11:58 -!- justanotheruser [~justanoth@gateway/tor-sasl/justanotheruser] has joined #hplusroadmap
12:56 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Ping timeout: 260 seconds]
13:24 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has joined #hplusroadmap
13:50 < hprmbridge> kanzure> https://writings.stephenwolfram.com/2024/05/why-does-biological-evolution-work-a-minimal-model-for-biological-evolution-and-other-adaptive-processes/
14:31 -!- Gooberpatrol66 [~Gooberpat@user/gooberpatrol66] has quit [Remote host closed the connection]
15:07 < hprmbridge> harmoniq.punk> It is clear that adaptive processes are working and are good to discover the lowest Kolmogorov complexity to solve a wide range of problems black boxed or not. The real question is "how do we invent a system, analogue to our brain that accelerate the adaptive process without having to wait years to evolve or autonomously learn something new?"  My answer is analogue signals. For decades we dream to
15:07 < hprmbridge> harmoniq.punk> "precision" so we invent digital signals, only to find ourselves today thriving for "approximation" forgetting that analog signals were and are still the state of the art in "approximation".  Drugs or plasmids discovery are adaptive process as well. To simulate new plasmid in digital domain you need billions of transistors while to simulate a plasmid in analog domain takes very few opamps,
15:07 < hprmbridge> harmoniq.punk> resistors and capacitors. The higher the analogue freq the higher bit depth is achievable which increase exponential with freq. Depending on the configuration and multiplexing, an analogue circuit in 500Ghz can perform as well all the GPUs in the world supposing they can be scaled horizontal all together. Our synapses operate in the Thz range of EM spectrum. Thz is 10^12 which is in the very far
15:07 < hprmbridge> harmoniq.punk> infrared spectrum. We know how to deal with parasitic inductance and capacitance @ 0.5 Thz and we can make it @ 1 Thz. The secret sauce is actually the circuit configuration, the mux/demux before AD/DA and the mathematical transformations. This cipset can be built as FPAA or dpASP even in 300nm or higher so with a bit of work can be done even in garage with DLP projector. Can cost under 10$ and
15:07 < hprmbridge> harmoniq.punk> doesn't need superconducting SF thermodynamic computation propose by Extropic to represent a huge distribution. 1Thz can have and enormous distribution using combinatorial modulation. Virtually inf entropy.
15:13 < L29Ah> digital computers are good for their versatility: certainly you can have effective and efficient analog computers for things like neural network inference passes or some massive modeling; the question is whether your scale can support the R&D required for the domain-specific hardware solution
15:14 < L29Ah> while software is relatively cheap, and the hardware for it is ubiquitous and liquid
15:14 < hprmbridge> harmoniq.punk> you don't need domain-specific hardware if you can configure from software. This is exactly like FPAA are doing
15:15 < L29Ah> yes, and we also have FPGAs for digital things, guess why they're restricted to quite narrow niches and don't appear in general use, even in modelling?
15:15 < hprmbridge> harmoniq.punk> and you don't need many spcific configurations, probably max 1000 primitives to solve some atractors
15:17 < hprmbridge> harmoniq.punk> FPGAs are literally everywhere, they are not narrow niches at all.
15:17 < L29Ah> pretty sure i have none in my home
15:18 < hprmbridge> harmoniq.punk> your TV, your computer mainboard, all your electronics
15:18 < L29Ah> perhaps we could design some relatively reusable analog processors with some drawbacks (that put them behind FPAAs/ASICs in some niches), and that would require an army of new kind of programmers to utilize effectively, the market attitude, etc...
15:18 < L29Ah> i don't have TV, my computer mainboard has no FPGAs, my electronics neither
15:19 < L29Ah> flash chips and PLDs are not FPGAs
15:20 < L29Ah> *PALs
15:23 < hprmbridge> harmoniq.punk> I'm not talkin about PLDs, many electronics have low price/low scale FPGAs to allow manufacturer flexibility. You don't need to have Altera and Xilinx. There are tons of Microsemi FPGAs used in all sorts of scenarios.
15:25 < hprmbridge> harmoniq.punk> Anyway, the point is that you can configure primitives from neural net framewrok
15:27 < L29Ah> perhaps they're used, but i don't see them because they don't go past low volume things or prototypes
15:28  * L29Ah hides
18:00 -!- darsie [~darsie@84-112-12-36.cable.dynamic.surfer.at] has quit [Ping timeout: 256 seconds]
19:00 -!- justanotheruser [~justanoth@gateway/tor-sasl/justanotheruser] has quit [Ping timeout: 260 seconds]
19:44 -!- justanotheruser [~justanoth@gateway/tor-sasl/justanotheruser] has joined #hplusroadmap
20:04 -!- TMM [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
20:04 -!- TMM [hp@amanda.tmm.cx] has joined #hplusroadmap
20:15 < nsh> Lynyrd Skynet - Sweet Home AI-labama - https://websim.ai/c/X8mtqCDhC1gQeB1RW
20:34 -!- mxz__ [~mxz@user/mxz] has joined #hplusroadmap
20:35 -!- mxz [~mxz@user/mxz] has quit [Ping timeout: 268 seconds]
20:35 -!- mxz__ is now known as mxz
20:36 -!- mxz_ [~mxz@user/mxz] has quit [Ping timeout: 256 seconds]
21:22 < fenn> "FPGAs for digital things, guess why they're restricted to quite narrow niches and don't appear in general use" <- because of the overbearing IP maximalist duopoly that has been strangling the scene for decades
22:11 -!- helleshin [~talinck@76-230-66-1.lightspeed.cntmoh.sbcglobal.net] has quit [Ping timeout: 268 seconds]
22:13 -!- helleshin [~talinck@76-230-66-1.lightspeed.cntmoh.sbcglobal.net] has joined #hplusroadmap
22:41 -!- mxz_ [~mxz@user/mxz] has joined #hplusroadmap
--- Log closed Mon May 06 00:00:47 2024