| --- Log opened Wed Mar 09 00:00:55 2016 | ||
| -!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-lnjycttzktwwicwk] has joined ##hplusroadmap | 00:32 | |
| -!- jtimon [~quassel@35.31.134.37.dynamic.jazztel.es] has quit [Ping timeout: 276 seconds] | 01:06 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has quit [Quit: Quit] | 01:12 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has joined ##hplusroadmap | 01:46 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has quit [Ping timeout: 240 seconds] | 03:20 | |
| -!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-lnjycttzktwwicwk] has quit [Quit: Connection closed for inactivity] | 03:50 | |
| -!- jaboja [~jaboja@eml127.neoplus.adsl.tpnet.pl] has joined ##hplusroadmap | 04:10 | |
| -!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has joined ##hplusroadmap | 04:11 | |
| -!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-jbnbuljizzdodtqy] has joined ##hplusroadmap | 04:20 | |
| -!- T0BI [~ObitO@41.101.163.255] has joined ##hplusroadmap | 04:25 | |
| -!- blueskin [~blueskin@unaffiliated/blueskin] has quit [Ping timeout: 240 seconds] | 04:27 | |
| -!- jron [~okok@ec2-54-161-129-226.compute-1.amazonaws.com] has quit [Ping timeout: 240 seconds] | 04:27 | |
| -!- TOB1 [~ObitO@41.101.163.255] has quit [Ping timeout: 240 seconds] | 04:27 | |
| -!- HEx1 [~HEx@hexwab.plus.com] has quit [Ping timeout: 240 seconds] | 04:27 | |
| -!- HEx1 [~HEx@hexwab.plus.com] has joined ##hplusroadmap | 04:27 | |
| -!- blueskin [~blueskin@unaffiliated/blueskin] has joined ##hplusroadmap | 04:32 | |
| -!- jron [~okok@ec2-54-161-129-226.compute-1.amazonaws.com] has joined ##hplusroadmap | 04:34 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has joined ##hplusroadmap | 04:37 | |
| -!- jtimon [~quassel@35.31.134.37.dynamic.jazztel.es] has joined ##hplusroadmap | 04:46 | |
| -!- eudoxia [~eudoxia@r167-56-157-202.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 05:08 | |
| -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-rdxopouqtaebteig] has joined ##hplusroadmap | 05:13 | |
| -!- jaboja [~jaboja@eml127.neoplus.adsl.tpnet.pl] has quit [Ping timeout: 268 seconds] | 05:29 | |
| -!- seanph [~seanph@104.237.91.209] has joined ##hplusroadmap | 05:47 | |
| -!- seanph [~seanph@104.237.91.209] has quit [Remote host closed the connection] | 05:48 | |
| -!- sandeepkr [~sandeep@111.235.64.4] has quit [Ping timeout: 276 seconds] | 05:49 | |
| -!- jaboja [~jaboja@eml127.neoplus.adsl.tpnet.pl] has joined ##hplusroadmap | 05:50 | |
| -!- seanph [~seanph@104.237.91.209] has joined ##hplusroadmap | 05:51 | |
| -!- T0BI [~ObitO@41.101.163.255] has quit [Ping timeout: 244 seconds] | 05:52 | |
| -!- T0BI [~ObitO@41.101.163.217] has joined ##hplusroadmap | 05:52 | |
| -!- AmbulatoryCortex [~Ambulator@173-31-155-69.client.mchsi.com] has joined ##hplusroadmap | 05:57 | |
| -!- seanph [~seanph@104.237.91.209] has quit [Remote host closed the connection] | 06:07 | |
| -!- sandeepkr [~sandeep@111.235.64.4] has joined ##hplusroadmap | 06:11 | |
| -!- jaboja [~jaboja@eml127.neoplus.adsl.tpnet.pl] has quit [Ping timeout: 248 seconds] | 06:38 | |
| -!- seanph [~seanph@104.237.91.102] has joined ##hplusroadmap | 07:07 | |
| -!- seanph [~seanph@104.237.91.102] has quit [Ping timeout: 248 seconds] | 07:12 | |
| kanzure | http://thebaffler.com/salvos/everybody-freeze-pein | 07:41 |
|---|---|---|
| -!- sandeepkr [~sandeep@111.235.64.4] has quit [Ping timeout: 268 seconds] | 07:44 | |
| -!- PatrickRobotham [uid18270@gateway/web/irccloud.com/x-jbnbuljizzdodtqy] has quit [Quit: Connection closed for inactivity] | 07:50 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has quit [Ping timeout: 268 seconds] | 07:51 | |
| eudoxia | >In Alcor’s account, “the actual success of perfusion in this case appears negligible.” (Perfusion is the term for pumping fluids through blood vessels.) A CT scan later confirmed that “cryoprotective perfusion was not generally successful”—meaning that Suozzi’s brain would not be well preserved. | 08:01 |
| eudoxia | damn | 08:01 |
| eudoxia | she was in a hospice in Scotsdale and everything and they *still* fucked up the perfusion? | 08:01 |
| eudoxia | it's not like she died of an aneurysm | 08:01 |
| eudoxia | >Credit goes to University of California–Berkeley lecturer in rhetoric Dale Carrico for this coinage. | 08:10 |
| eudoxia | aaaaaaaaaa | 08:10 |
| JayDugger | Yeah, citing Dale Carrico? | 08:12 |
| fenn | in suozzi's case, she was kicked out of the hospice for not dying fast enough, and then at the new place the night nurse shift changed right when she did die, so the alcor team wasn't able to enter until several hours later | 08:12 |
| eudoxia | :( | 08:13 |
| fenn | the second nurse got lost or something on the way to the hotel, and alcor needs a pronouncement of death before they can do anything | 08:13 |
| eudoxia | what a bunch of overpaid hacks | 08:14 |
| JayDugger | Kind of spoiled for choice of referents with your "bunch of overpaid hacks," eudoxia. | 08:16 |
| -!- sandeepkr [~sandeep@111.235.64.4] has joined ##hplusroadmap | 08:16 | |
| eudoxia | the hospice nurses | 08:16 |
| eudoxia | alcor should run an ez die hospice as a side business | 08:17 |
| eudoxia | just pay someone to sit by the patient until they stop breathing, how hard is that | 08:17 |
| JayDugger | That part, not very. Legal compliance? I doubt there's an upper limit for that part's difficulty. | 08:18 |
| eudoxia | hmm | 08:18 |
| JayDugger | I am so glad I had the cryonics talk with my wife before proposing. | 08:18 |
| JayDugger | Now, if I could just get her to entertain the idea of someday emigrating to Mars… | 08:20 |
| -!- sandeepkr [~sandeep@111.235.64.4] has quit [Ping timeout: 244 seconds] | 08:21 | |
| Aurelius_Work2 | JayDugger : my gf is down with all that shit. | 08:24 |
| Aurelius_Work2 | JayDugger : though the 'going to space' part itself might terrify her | 08:24 |
| JayDugger | Such are hard to find. Consider marriage. | 08:26 |
| kanzure | what? dn't give bad advice come on... | 08:27 |
| JayDugger | "Consider," not "do it now, where are my grandchildren, I can still disinherit you, etc." | 08:28 |
| Aurelius_Work2 | JayDugger : it's on the list | 08:28 |
| Aurelius_Work2 | we'll also be renegotiating a prenup yearly | 08:28 |
| JayDugger | Though yeah, if it doesn't work out, well, you were warned. :) | 08:29 |
| Aurelius_Work2 | vov | 08:29 |
| -!- bsm117532 [~mcelrath@38.121.165.30] has quit [Remote host closed the connection] | 08:42 | |
| fenn | mcelrath.. didn't he write about the bitcoin braid idea? | 08:49 |
| fenn | i guess it shouldn't be surprising | 08:51 |
| kanzure | he used to be extropian or something | 08:56 |
| kanzure | turns out all the bitcoiners are transhumanist, *shrug* | 08:56 |
| -!- nmz787_i [~ntmccork@134.134.139.77] has joined ##hplusroadmap | 09:03 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has joined ##hplusroadmap | 09:07 | |
| -!- jdqx [uid146399@gateway/web/irccloud.com/x-oswvgoflqocqiqnl] has joined ##hplusroadmap | 09:10 | |
| -!- bsm117532 [~mcelrath@38.121.165.30] has joined ##hplusroadmap | 09:33 | |
| kanzure | http://www.nature.com/news/open-hardware-pioneers-push-for-low-cost-lab-kit-1.19518 | 10:03 |
| -!- seanph [~seanph@104.237.91.102] has joined ##hplusroadmap | 10:14 | |
| -!- eudoxia_ [~eudoxia@r167-56-34-224.dialup.adsl.anteldata.net.uy] has joined ##hplusroadmap | 10:20 | |
| -!- eudoxia [~eudoxia@r167-56-157-202.dialup.adsl.anteldata.net.uy] has quit [Read error: Connection reset by peer] | 10:20 | |
| -!- seanph [~seanph@104.237.91.102] has quit [Ping timeout: 248 seconds] | 10:20 | |
| -!- eudoxia_ [~eudoxia@r167-56-34-224.dialup.adsl.anteldata.net.uy] has quit [Quit: Leaving] | 10:25 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has joined ##hplusroadmap | 10:28 | |
| -!- jcluck is now known as cluckj | 10:31 | |
| -!- ArturShaik [~ArturShai@37.218.141.136] has quit [Ping timeout: 252 seconds] | 10:58 | |
| nmz787_i | zopr | 11:18 |
| nmz787_i | zorp | 11:18 |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has quit [Ping timeout: 260 seconds] | 11:28 | |
| -!- Aurelius_Work2 [~cpopell@209.48.69.2] has quit [Ping timeout: 252 seconds] | 11:41 | |
| -!- jaboja [~jaboja@eml127.neoplus.adsl.tpnet.pl] has joined ##hplusroadmap | 11:45 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has joined ##hplusroadmap | 11:46 | |
| -!- jaboja64 [~jaboja@ejr220.neoplus.adsl.tpnet.pl] has joined ##hplusroadmap | 12:06 | |
| -!- jaboja [~jaboja@eml127.neoplus.adsl.tpnet.pl] has quit [Ping timeout: 264 seconds] | 12:10 | |
| nmz787_i | bing bong | 12:13 |
| c0rw1n | bong bing | 12:15 |
| nmz787_i | microsoft bing | 12:20 |
| nmz787_i | /me I've been tricked! | 12:20 |
| * harold hugs nmz787_i | 12:23 | |
| -!- Filosofem [~Jawmare@unaffiliated/jawmare] has joined ##hplusroadmap | 12:39 | |
| -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-rdxopouqtaebteig] has quit [Quit: Connection closed for inactivity] | 12:41 | |
| -!- Jawmare [~Jawmare@unaffiliated/jawmare] has quit [Ping timeout: 276 seconds] | 12:42 | |
| -!- Filosofem [~Jawmare@unaffiliated/jawmare] has quit [Read error: Connection reset by peer] | 12:55 | |
| -!- Filosofem [~Jawmare@unaffiliated/jawmare] has joined ##hplusroadmap | 12:56 | |
| -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-lxmnykroovtlaxas] has joined ##hplusroadmap | 13:00 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has quit [Ping timeout: 244 seconds] | 13:30 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has joined ##hplusroadmap | 13:40 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has quit [Ping timeout: 244 seconds] | 13:59 | |
| -!- T0BI [~ObitO@41.101.163.217] has quit [] | 14:00 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has joined ##hplusroadmap | 14:02 | |
| -!- jaboja64 [~jaboja@ejr220.neoplus.adsl.tpnet.pl] has quit [Ping timeout: 264 seconds] | 14:24 | |
| -!- nsh [~lol@wikipedia/nsh] has quit [Quit: ZNC - http://znc.in] | 14:28 | |
| -!- nsh [~lol@wikipedia/nsh] has joined ##hplusroadmap | 14:31 | |
| -!- nmz787_i [~ntmccork@134.134.139.77] has quit [Quit: Leaving.] | 14:44 | |
| -!- nmz787_i [~ntmccork@134.134.139.74] has joined ##hplusroadmap | 14:59 | |
| -!- berndj [~berndj@azna.co.za] has quit [Ping timeout: 268 seconds] | 15:01 | |
| -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-lxmnykroovtlaxas] has quit [Quit: Connection closed for inactivity] | 15:01 | |
| -!- amiller [~socrates1@unaffiliated/socrates1024] has quit [Ping timeout: 268 seconds] | 15:01 | |
| -!- berndj [~berndj@azna.co.za] has joined ##hplusroadmap | 15:03 | |
| -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-zqvdckqnjinularg] has joined ##hplusroadmap | 15:04 | |
| -!- nmz787_i1 [ntmccork@nat/intel/x-sbwfvujxkluhcmkh] has joined ##hplusroadmap | 15:04 | |
| -!- nmz787_i [~ntmccork@134.134.139.74] has quit [Ping timeout: 244 seconds] | 15:04 | |
| -!- nmz787_i [~ntmccork@134.134.139.74] has joined ##hplusroadmap | 15:05 | |
| -!- nmz787_i [~ntmccork@134.134.139.74] has quit [Client Quit] | 15:07 | |
| -!- nmz787_i1 [ntmccork@nat/intel/x-sbwfvujxkluhcmkh] has quit [Ping timeout: 276 seconds] | 15:09 | |
| -!- Guest88865 [~socrates1@li175-104.members.linode.com] has joined ##hplusroadmap | 15:11 | |
| -!- Guest88865 [~socrates1@li175-104.members.linode.com] has quit [Ping timeout: 268 seconds] | 15:16 | |
| -!- amiller_ [~socrates1@li175-104.members.linode.com] has joined ##hplusroadmap | 15:21 | |
| -!- bsm117532 [~mcelrath@38.121.165.30] has quit [Remote host closed the connection] | 15:25 | |
| -!- maaku [180460d5@gateway/web/freenode/ip.24.4.96.213] has joined ##hplusroadmap | 15:44 | |
| maaku | I am taking much schadenfreude in how lesswrong reacts to AlphaGo defeating Lee Sedol | 15:45 |
| kanzure | links? | 15:46 |
| kanzure | or do you mean on irc | 15:46 |
| kanzure | http://people.idsia.ch/~juergen/ | 15:47 |
| maaku | re: the match https://gogameguru.com/alphago-defeats-lee-sedol-game-1/ , lw: http://lesswrong.com/r/discussion/lw/ne1/alphago_versus_lee_sedol/ | 15:47 |
| -!- mf1008 [~mf1008@unaffiliated/mf1008] has quit [Ping timeout: 246 seconds] | 15:54 | |
| -!- mf1008 [~mf1008@unaffiliated/mf1008] has joined ##hplusroadmap | 15:55 | |
| maaku | but I suspect the wise stewards at MIRI will find some way to reason themselves into believing that defeating humans at Go says nothing about the advancement of AGI | 15:55 |
| -!- a7b9d6e6 [~a7b9d6e6@static.214.43.251.148.clients.your-server.de] has joined ##hplusroadmap | 15:57 | |
| -!- a7b9d6e6_ [~a7b9d6e6@static.214.43.251.148.clients.your-server.de] has quit [Ping timeout: 252 seconds] | 15:59 | |
| -!- maaku [180460d5@gateway/web/freenode/ip.24.4.96.213] has quit [Quit: Page closed] | 16:00 | |
| -!- chris_99 [~chris_99@unaffiliated/chris-99/x-3062929] has quit [Quit: Leaving] | 16:00 | |
| Houshalter | maaku, um what? Yudkowsky was saying a month ago that beating go was a significant advancement towards AGI. That alphago was a sign that AGI might be close | 16:11 |
| Houshalter | EY left comments on the HN discussion about it. and the #lesswrong IRC seems to be really excited about the alphago victory | 16:12 |
| -!- seanph [~seanph@104.237.91.102] has joined ##hplusroadmap | 16:14 | |
| -!- psyreal [~psyreal@mail.transworldhealth.com] has joined ##hplusroadmap | 16:18 | |
| -!- seanph [~seanph@104.237.91.102] has quit [Ping timeout: 240 seconds] | 16:19 | |
| -!- fleshtheworld [~fleshthew@2602:306:cf0f:4c20:71de:2189:a092:f140] has quit [Read error: Connection reset by peer] | 16:19 | |
| -!- AmbulatoryC0rtex [~Ambulator@173-31-155-69.client.mchsi.com] has joined ##hplusroadmap | 16:43 | |
| -!- strages_ [sid11297@gateway/web/irccloud.com/x-kkhtfrhsjlmonyuh] has joined ##hplusroadmap | 16:46 | |
| -!- juri__ [~juri@vpn166.sdf.org] has joined ##hplusroadmap | 16:48 | |
| -!- strages [sid11297@gateway/web/irccloud.com/x-lfbrxpvmzhyypklb] has quit [Ping timeout: 248 seconds] | 16:49 | |
| -!- juri_ [~juri@vpn166.sdf.org] has quit [Ping timeout: 248 seconds] | 16:49 | |
| -!- AmbulatoryCortex [~Ambulator@173-31-155-69.client.mchsi.com] has quit [Ping timeout: 248 seconds] | 16:49 | |
| -!- HEx1 [~HEx@hexwab.plus.com] has quit [Ping timeout: 248 seconds] | 16:49 | |
| -!- juri__ is now known as juri_ | 16:49 | |
| -!- HEx1 [~HEx@hexwab.plus.com] has joined ##hplusroadmap | 16:50 | |
| -!- strages_ is now known as strages | 16:50 | |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has quit [Ping timeout: 260 seconds] | 17:05 | |
| -!- maaku [~quassel@173-228-107-141.dsl.static.fusionbroadband.com] has joined ##hplusroadmap | 17:44 | |
| -!- drewbot [~cinch@ec2-54-159-179-143.compute-1.amazonaws.com] has quit [Remote host closed the connection] | 17:45 | |
| -!- drewbot [~cinch@ec2-54-158-133-117.compute-1.amazonaws.com] has joined ##hplusroadmap | 17:45 | |
| -!- AmbulatoryC0rtex [~Ambulator@173-31-155-69.client.mchsi.com] has quit [Read error: Connection reset by peer] | 18:07 | |
| Taek | I still don't see how this moves us towards general intelligence. Is the computer capable of performing an original thought without brute-force? Or is it the case that the human brain does some variation of brute force? | 18:10 |
| Taek | hmm. I prefer to convince myself that it's still a dumb machine in the same way a chess engine is dumb than have to face the idea that there's a greater intelligence out there | 18:10 |
| Taek | I don't think we're ready to handle the changes it would herald | 18:11 |
| Houshalter | Taek, the thing is it's not a brute force game playing engine. The search space of Go is far too large to use those techniques. It uses neural networks, which learn to recognize patterns and learn strategies. The same way humans play. And it's a very general algorithm, that can be applied to many domains and games | 18:19 |
| -!- yashgaroth [~yashgarot@2602:306:35fa:d500:f5e0:f867:a11d:8d52] has joined ##hplusroadmap | 18:20 | |
| Taek | Houshalter: it had to play itself billions of times. The pattern matching arose from trying things many times and determining which ones were successful. I certainly didn't learn chess by playing billions of imaginary games. (I don't play go) | 18:20 |
| Taek | I do acknowledge that it's a huge improvement over chess | 18:20 |
| Taek | As long as it's not demonstrating sentience, I'm not really worried | 18:21 |
| Taek | then it's just a tool to find better drugs, etc. | 18:21 |
| Taek | though, if Google is the only one with that tech, they become utterly dominant | 18:21 |
| kanzure | Taek: part of the problem is that thinking is underspecified | 18:21 |
| kanzure | and sentience is also underspecified | 18:22 |
| Taek | kanzure: fully acknowledged. I'm not sure I would be able to tell an AGI even if it had an unbounded conversation with me | 18:22 |
| kanzure | in fact i'm not really sure if the human brain is "thinking"- if "thinking" is underspecified then how would i ever be capable of being sure that the human brain does it? | 18:22 |
| -!- esmerelda [~andares@unaffiliated/jacco] has quit [Ping timeout: 268 seconds] | 18:22 | |
| kanzure | i agree that human brains seem to have some sort of cognitive ability information processing stuffs that computers do not seem to be doing yet | 18:22 |
| maaku | Task except that is how you play chess isn't it? You go through simulations in your head when you think out your next move. | 18:28 |
| maaku | I think AlphaGo is very close to how humans process and learn strategic games (based on self reports, not empirical neurobiology) | 18:29 |
| -!- Ferrous [~Ferrous@pool-71-119-223-183.lsanca.fios.verizon.net] has quit [Ping timeout: 260 seconds] | 18:30 | |
| kanzure | i usually play games by doing internal simulation of moves | 18:32 |
| maaku | That's certainly how I play chess | 18:32 |
| kanzure | you don't do random moves?? bah | 18:33 |
| kanzure | you have to introduce entropy, that's how you throw the ai off | 18:33 |
| kanzure | :P | 18:33 |
| -!- psyreal [~psyreal@mail.transworldhealth.com] has quit [Ping timeout: 244 seconds] | 18:33 | |
| kanzure | the problem with self-reports is that i'm never quite sure if they map back to reality | 18:36 |
| maaku | What AlphaGo does that is both revolutionary and at the same time blindingly obvious in retrospect is self-modify using reinforcement learning over the output of those throwaway next move simulations | 18:36 |
| kanzure | yes, "have layers send feedback to other layers" and such | 18:37 |
| pasky | Taek: not billions, millions | 18:40 |
| kanzure | Taek: i think you would be able to detect an AGI by talking with it and then reasonably checking its claims, like if it's an author then where are all the things it has authored? where are the blog posts? etc. | 18:41 |
| maaku | Turing test? | 18:42 |
| kanzure | i think the traditional assumption about "turing test" is that you don't have an internet connection, or something | 18:44 |
| Houshalter | Taek, eh, not entirely true. We don't know how many times it played against itself, but in the match it took 2 minutes to make a move. And it only had 5 months or so to train. That would be thousands of times, which is about as many games as a human professional has played in their lifetime | 18:44 |
| -!- justanotheruser [~Justan@unaffiliated/justanotheruser] has quit [Ping timeout: 260 seconds] | 18:45 | |
| Houshalter | however the ability of computers to learn things much more rapidly than humans is a huge advantage in itself | 18:45 |
| -!- justanotheruser [~Justan@unaffiliated/justanotheruser] has joined ##hplusroadmap | 18:46 | |
| Taek | maaku: when I play chess I look for specific patterns. Is my queen and bishop lined up? Is the queen in front? Are my pawns stacked like X or Y. I'm only a 1400 player or so, but I find that I play better when I focus on keeping my side of the board 'clean', which is definitely a layer beyond brute-force or pure tactics. | 18:47 |
| Taek | I brute force on top of that, but I have heavy heruistics to evaluate the strength of a given position, chess enginges typically make those evaluations by brute-forcing really deep, and not using too many heuristics | 18:48 |
| Taek | deep mind is of course a different beast | 18:48 |
| Houshalter | Taek, NNs are not AGI, but they are step towards AGI. They can learn very similar to the way animal brains learn. There will be have to be many more insights and developments before they get to human level, but I believe that will happen. In the meantime they can still replace like 90% of human's jobs | 18:50 |
| Houshalter | maaku, you missed my reply earlier. MIRI is not dismissing alphago. in fact Yudkowsky has been saying for the last month that it represents significant progress in AI. | 18:51 |
| -!- |node [uid125132@gateway/web/irccloud.com/x-pqzkndeyruhmxhry] has joined ##hplusroadmap | 18:56 | |
| -!- |node [uid125132@gateway/web/irccloud.com/x-pqzkndeyruhmxhry] has quit [Client Quit] | 18:58 | |
| -!- jdqx [uid146399@gateway/web/irccloud.com/x-oswvgoflqocqiqnl] has quit [Quit: Connection closed for inactivity] | 19:16 | |
| -!- sandeepkr [~sandeep@111.235.64.4] has joined ##hplusroadmap | 19:20 | |
| -!- wrldpcmbl [uid145438@gateway/web/irccloud.com/x-djpztnwfpuuglrhu] has joined ##hplusroadmap | 19:39 | |
| -!- ArturShaik [~ArturShai@37.218.141.136] has joined ##hplusroadmap | 19:55 | |
| kanzure | http://www.nature.com/news/governance-learn-from-diy-biologists-1.19507 | 20:01 |
| -!- CheckDavid [uid14990@gateway/web/irccloud.com/x-zqvdckqnjinularg] has quit [Quit: Connection closed for inactivity] | 20:01 | |
| -!- sandeepkr [~sandeep@111.235.64.4] has quit [Ping timeout: 244 seconds] | 20:12 | |
| maaku | Houshalter I have no idea what EY has written post LW other than HPMoR because he moved to the closed medium of Facebook. | 20:14 |
| maaku | I know that a big portion of the sequences was spent rejecting NN as a messy study of emergent behavior and therefore trivially wrong | 20:16 |
| maaku | If he has moved beyond that, of is good. | 20:16 |
| kanzure | i suppose "NNs are not sufficiently studied" is a good criticism | 20:17 |
| Houshalter | maaku, he wrote on facebook, and he's been making comments on HN about it recently | 20:17 |
| maaku | However my snarky vindicative pleasure is from AGI being around the corner rather than the party line of 50+ years out | 20:17 |
| maaku | kanzure I believe the true rejection was that successful NN have been "not Bayesian" whatever that means | 20:18 |
| kanzure | oh that's just cfar "church of bayes" cult stuff | 20:19 |
| maaku | Task -- AlphaGo is using the same type of strategies | 20:21 |
| maaku | Taek | 20:21 |
| maaku | It is training neural networks to recognize such patterns and appropriate responses. | 20:22 |
| Houshalter | the thing is game playing doesn't require bayes at all. Bayes is great about protecting about overfitting. But alphago can self play, or look at millions of games by humans. It doesn't need to worry about overfitting too much | 20:23 |
| Houshalter | bayes is great, but not magic AI sauce | 20:23 |
| -!- sandeepkr [~sandeep@111.235.64.4] has joined ##hplusroadmap | 20:52 | |
| fenn | maaku the AI worriers i know were significantly more worried after hearing about alphago | 21:07 |
| -!- psyreal [~psyreal@71.92.165.184] has joined ##hplusroadmap | 21:13 | |
| -!- Netsplit *.net <-> *.split quits: wrldpcmbl, yashgaroth | 21:40 | |
| -!- Netsplit over, joins: yashgaroth | 21:46 | |
| -!- Netsplit *.net <-> *.split quits: yashgaroth | 22:03 | |
| -!- Netsplit over, joins: yashgaroth | 22:08 | |
| -!- Houshalter [~Houshalte@oh-71-50-63-56.dhcp.embarqhsd.net] has quit [Quit: Quit] | 22:18 | |
| kanzure | he would know, fenn lives at "ai worry" hq | 22:24 |
| -!- Gurkenglas [Gurkenglas@dslb-188-106-113-006.188.106.pools.vodafone-ip.de] has joined ##hplusroadmap | 22:34 | |
| maaku | fenn - if it prods them to work in a practical and useful direction, I'd be very happy | 22:36 |
| maaku | Houshalter what I don't understand is how "Bayesian reasoning" is any different than regular old probability theory that underlies 98% of the AI work out there, NARS being the (only?) exception | 22:40 |
| maaku | Probability is probability, regardless of whether you describe things as priors and posteriors or some other formulation | 22:43 |
| maaku | And yeah, not magic AI sauce indeed | 22:43 |
| -!- jdqx [uid146399@gateway/web/irccloud.com/x-nmbydlureaprwyuv] has joined ##hplusroadmap | 22:50 | |
| -!- yashgaroth [~yashgarot@2602:306:35fa:d500:f5e0:f867:a11d:8d52] has quit [Quit: Leaving] | 22:56 | |
| -!- psyreal [~psyreal@71.92.165.184] has quit [Ping timeout: 246 seconds] | 23:23 | |
| --- Log closed Thu Mar 10 00:00:55 2016 | ||
Generated by irclog2html.py 2.15.0.dev0 by Marius Gedminas - find it at mg.pov.lt!