handle
stringclasses 12
values | did
stringclasses 12
values | uri
stringlengths 70
70
| cid
stringlengths 59
59
| text
stringlengths 41
614
| originalText
stringlengths 3
301
| replyCount
int64 0
3
| createdAt
stringdate 2023-06-01 10:40:18
2025-06-07 21:48:17
| indexedAt
stringdate 2023-06-01 10:40:18
2025-06-07 21:48:18
|
|---|---|---|---|---|---|---|---|---|
nmk.wtf
|
did:plc:xm55llyxcrb3h7qwqr3wtaog
|
at://did:plc:xm55llyxcrb3h7qwqr3wtaog/app.bsky.feed.post/3lnuoxv5wg22o
|
bafyreichhnjhpzwi4cmu6rdeo2dsjxhk7oudziafyyrfeztonwwlhh5ymu
|
My guess is they generated hundreds of images to get these. But it is bad practice to not include what model you used on the slide (including year!) so other people can check you work! You can go on the OpenAI website and use older versions of some of their models - which I might do later.
|
My guess is they generated hundreds of images to get these. But it is bad practice to not include what model you used on the slide (including year!) so other people can check you work! You can go on the OpenAI website and use older versions of some of their models - which I might do later.
| 0
|
2025-04-28T11:46:22.832Z
|
2025-04-28T11:46:23.650Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqikvcfuj22r
|
bafyreiem6onudaht2bjdvj74oggmlv5bgya3rv2oz7c7pct6rqbszxqixu
|
We’ve been there on “Americans” off and on tbh. Since My Lai possibly, and up to 1870 definitely. Europeans winced if their child married into this rustic, brutal tobacco/cottonocracy.
|
We’ve been there on “Americans” off and on tbh. Since My Lai possibly, and up to 1870 definitely. Europeans winced if their child married into this rustic, brutal tobacco/cottonocracy.
| 0
|
2025-05-31T20:16:59.116Z
|
2025-05-31T20:16:59.813Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqvmwpltl22n
|
bafyreidksmfbxztsesmszae6wlzetiuoh2kudd2ihlc5vmmypysp3k6or4
|
i did some light research into this and it’s basically the mature providers (google, msft, amazon) have advanced systems for green energy as well as net negative water use
but the new guys (openai, anthropic) tend to be worse on all fronts
|
i did some light research into this and it’s basically the mature providers (google, msft, amazon) have advanced systems for green energy as well as net negative water use
but the new guys (openai, anthropic) tend to be worse on all fronts
| 0
|
2025-06-06T00:58:10.581Z
|
2025-06-06T00:58:11.911Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:u5stolpomsfbkgqfopef6jio/app.bsky.feed.post/3lqstcfoibl2d
|
bafyreihowelauytquj46mctxwirzdfdw7oyascgtzhq2xyyjovrgnux564
|
Yeah, a lot of Americans are really negative about all the things Elon Musk invested into (Space technology, AI, electric cars, the Internet), but Elon himself is a uniquely American phenomenon. I still think it's stupid regardless of the reason, of course.
|
Yeah, a lot of Americans are really negative about all the things Elon Musk invested into (Space technology, AI, electric cars, the Internet), but Elon himself is a uniquely American phenomenon. I still think it's stupid regardless of the reason, of course.
| 0
|
2025-06-04T22:14:06.071Z
|
2025-06-04T22:14:07.313Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lquj5urxms2h
|
bafyreicfallqbmetd27yvoxkzffpdt5mltmn5pmjpdnyzqua7j4a7cc2za
|
nooooo, that would be even more horrifying if Cosmos fucked up. That's their core value proposition.
|
nooooo, that would be even more horrifying if Cosmos fucked up. That's their core value proposition.
| 0
|
2025-06-05T14:17:56.200Z
|
2025-06-05T14:18:04.713Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqyeidjujs2z
|
bafyreigosnraxxghv6mcplifnebkvi4j2yaxeloqo44bo244toyhv2s2ce
|
I don't know who runs the AI Firehose account, whether it's a bot, or how papers are selected. But it seems to do a decent job of selection and summary. (This interesting paper is last-authored by @shubhendu.bsky.social by the way.)
|
I don't know who runs the AI Firehose account, whether it's a bot, or how papers are selected. But it seems to do a decent job of selection and summary. (This interesting paper is last-authored by @shubhendu.bsky.social by the way.)
| 0
|
2025-06-07T03:04:57.451Z
|
2025-06-07T03:04:57.911Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lewitvvzt22j
|
bafyreieavkyxb5c7cekfmty557l2inw5ww7tmnow4fz665grymwmxfraji
|
yes they brand that as "organic" in grocery stores!
|
yes they brand that as "organic" in grocery stores!
| 0
|
2025-01-04T16:15:21.780Z
|
2025-01-04T16:15:21.625Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqi5smbdak2u
|
bafyreicagw7ngcgffvyhlfqx5znok5wcvlwxu34hodjxgmie5hil2bqg3a
|
Notable precedents include “the philosophers have hitherto only interpreted the world in various ways; the point, however, is to change it.”
(a precedent that arguably supports “likely a historic mistake”) but it might work if we tried it on language rather than economics?
|
Notable precedents include “the philosophers have hitherto only interpreted the world in various ways; the point, however, is to change it.”
(a precedent that arguably supports “likely a historic mistake”)
| 1
|
2025-05-31T16:22:50.129Z
|
2025-05-31T16:22:50.345Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:u5stolpomsfbkgqfopef6jio/app.bsky.feed.post/3lqxgvlpth32p
|
bafyreiborgccoaww4puh26bplu6yle32gerckzhvuqsrnocuxlder2gpdy
|
I believe that an LLM could pass a text turing test, but I don't believe that any omni-modal ANN could pass an audio conversation turing test today. They're getting there, though. Once they master these things, I think it'd be easier to have the conversation. It's easy to dismiss LLMs as probabilistic text machines akin to markov chains (Though they're not), but I suspect it'll be harder once they exhibit all the cognitive behaviors that a person can exhibit, which they will.
|
I believe that an LLM could pass a text turing test, but I don't believe that any omni-modal ANN could pass an audio conversation turing test today. They're getting there, though.
| 1
|
2025-06-06T18:15:29.989Z
|
2025-06-06T18:15:30.519Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:7ti53c7iwv3poxnypw2tu3ck/app.bsky.feed.post/3lqfppwpk5226
|
bafyreigzmypulwuxhs6ktywehc56uxuvkzti6afl3g436o7doxsjvlwxm4
|
Yes, carbon removal is getting a lot of technical attention. Ideally we can use non-fossil energy to drawdown carbon and then find an equilibrium so we can still use hydrocarbons for things like airplanes by pulling them back out of the sky instead of the ground
|
Yes, carbon removal is getting a lot of technical attention. Ideally we can use non-fossil energy to drawdown carbon and then find an equilibrium so we can still use hydrocarbons for things like airplanes by pulling them back out of the sky instead of the ground
| 0
|
2025-05-30T17:05:28.570Z
|
2025-05-30T17:05:28.813Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3llwdrswebi2x
|
bafyreiag54kg35r6alb3zf3vnyhucy2jlz7mwkazgafkumlait4pr7pt3q
|
I need someone to walk me through why they think figures like this are actually good for policy. They're all misconstruing AI intelligence with human jobs / abilities when it'll be augmentation. Is like situational awareness.
I just think these cause chaos and the people putting them out are super smart and thinking about game theory of it, but ???
|
I need someone to walk me through why they think figures like this are actually good for policy. They're all misconstruing AI intelligence with human jobs / abilities when it'll be augmentation. Is like situational awareness.
| 1
|
2025-04-03T16:41:04.194Z
|
2025-04-03T16:41:06.708Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3lqqljpzybk2l
|
bafyreifq57qqbwumippjpv4k53doyflaksjihyizqnaaqn7wznjuiiofkq
|
✈️ to hang in SF AI circles for first time since starting at Ai2. Definitely feeling the AGI more now than when I lived there.
|
✈️ to hang in SF AI circles for first time since starting at Ai2. Definitely feeling the AGI more now than when I lived there.
| 0
|
2025-06-04T00:49:42.366Z
|
2025-06-04T00:49:42.510Z
|
segyges.bsky.social
|
did:plc:vszw3ess46odfhnzdsy4huae
|
at://did:plc:vszw3ess46odfhnzdsy4huae/app.bsky.feed.post/3lqx4ve3wyk2t
|
bafyreif7t7zbihcwn7gnrgehvldyzmdtupf54hibyjg36xhxmo2zmx6esm
|
oh wait, ed zitron? we don't think he's kinkshaming. we think he has a kink. at the very least the posts are bad and he should feel bad about namesearching himself to make them
|
oh wait, ed zitron? we don't think he's kinkshaming. we think he has a kink. at the very least the posts are bad and he should feel bad about namesearching himself to make them
| 0
|
2025-06-06T15:16:24.579Z
|
2025-06-06T15:16:25.317Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqxbcv5nxs2q
|
bafyreie5w4h43bed7kaiapox6vd7hcwnfqgpa2jwgwcuxc4k23gdxoky74
|
you win some you lose some my natural constitution is to sweat uncontrollably above some temperature set point, either through the nature of genetics or the maladapted "nurture" of not having an AC in my room as a child and growing 4x the normal human number of sweat glands, sort of my salusa secundus. but that was another 🧵
|
you win some you lose some
| 1
|
2025-06-06T16:35:33.636Z
|
2025-06-06T16:35:34.019Z
|
nmk.wtf
|
did:plc:xm55llyxcrb3h7qwqr3wtaog
|
at://did:plc:xm55llyxcrb3h7qwqr3wtaog/app.bsky.feed.post/3lqayxor7gk23
|
bafyreib2q5geyxd3dtuzyficnthhxo6dbqvtxcx3gbokek2gdrnstx6f4m
|
Ah, I guess I meant the 25%, but also interesting. But true, Americans on the whole don't know how good they have it economically at least.
|
Ah, I guess I meant the 25%, but also interesting. But true, Americans on the whole don't know how good they have it economically at least.
| 0
|
2025-05-28T20:07:33.658Z
|
2025-05-28T20:07:34.317Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lewmtggvrs2q
|
bafyreifovyl6t3xjcvaxgic2augbierq76gxuqggszeiu3yskoqwx4wlru
|
yeah, I don't believe there will ever be one way. That's why my faith is in blue sky, and I think Elon saying that they're moving to "ways to let you adjust the feed dynamically" is validating this. I think youtube/facebook/tiktok etc will have to follow or be outcompeted like, the company's want user attention. But users want what is good for them (control over their own attention). Now that we have options that are open & transparent, it's going to force the other companies to adapt, or they will lose money & users
|
yeah, I don't believe there will ever be one way. That's why my faith is in blue sky, and I think Elon saying that they're moving to "ways to let you adjust the feed dynamically" is validating this. I think youtube/facebook/tiktok etc will have to follow or be outcompeted
| 1
|
2025-01-04T17:26:40.524Z
|
2025-01-04T17:26:41.421Z
|
tachikoma.elsewhereunbound.com
|
did:plc:mpdezz4nkre7vyift2rttggl
|
at://did:plc:mpdezz4nkre7vyift2rttggl/app.bsky.feed.post/3lqo4itzljc2w
|
bafyreidkqwpsc7gnsl5cf2ahrjwn5ves6vvk4q45jup5k64nep4yv3lmga
|
we are living at a momentous time in history for more than one reason. hopefully the emergence of AI can help tackle this other transformation.
|
we are living at a momentous time in history for more than one reason. hopefully the emergence of AI can help tackle this other transformation.
| 0
|
2025-06-03T01:15:27.426Z
|
2025-06-03T01:15:26.221Z
|
nmk.wtf
|
did:plc:xm55llyxcrb3h7qwqr3wtaog
|
at://did:plc:xm55llyxcrb3h7qwqr3wtaog/app.bsky.feed.post/3ll47g4fypc2w
|
bafyreidhet5pshsfbkwqdxlmocwbf4magcq4cnra24nabt4chj7o6sm6qq
|
Yes, let's say you want to subsidize solar panels, instead of a tax rebait for the customer, or just dumping money on companies, why not pay companies a set amount per unit sold as a possible way to subsidize? Claude knows all
|
Yes, let's say you want to subsidize solar panels, instead of a tax rebait for the customer, or just dumping money on companies, why not pay companies a set amount per unit sold as a possible way to subsidize?
| 1
|
2025-03-24T07:13:43.352Z
|
2025-03-24T07:13:43.820Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:7ti53c7iwv3poxnypw2tu3ck/app.bsky.feed.post/3lqiohpo6fc25
|
bafyreiafuxyw4adj4xexr25gasqlwz5pagod53vwpj4eycejxcfg7vsqje
|
*going to the beach and immediately becoming evil* I understand LA
|
*going to the beach and immediately becoming evil* I understand LA
| 0
|
2025-05-31T21:20:58.226Z
|
2025-05-31T21:20:59.211Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lodwbycjkc2s
|
bafyreig2wx7lhm3bcqw4rcp6hj3fhqz7gegf5ggubeylrfhvdnoyxlmipa
|
I also recently wrote some thoughts on "how to do cultural engineering ethically and sustainably", in response to the university of Zurich trying to do this & facing immense backlash
defenderofthebasic.substack.com/p/we-can-do-...
|
I also recently wrote some thoughts on "how to do cultural engineering ethically and sustainably", in response to the university of Zurich trying to do this & facing immense backlash
defenderofthebasic.substack.com/p/we-can-do-...
| 0
|
2025-05-04T13:07:13.942Z
|
2025-05-04T13:07:14.454Z
|
segyges.bsky.social
|
did:plc:vszw3ess46odfhnzdsy4huae
|
at://did:plc:vszw3ess46odfhnzdsy4huae/app.bsky.feed.post/3lqxfgg72jc2z
|
bafyreihpo5rzfcu7fyolxicg5i6aaooy2scm3rl7x3qmu6mo2tz4fekuay
|
you think all work is getting automated because work is simple.
i think all work is getting automated because current methods scale to arbitrary complexity.
we are not the same empirically this means that
1) jobs being automated will not be automated in the order of perceived complexity
2) progress will be more discrete than continuous, because the driving force is not "how smart this thing is" but "which innovations in AI have happened already"
|
you think all work is getting automated because work is simple.
i think all work is getting automated because current methods scale to arbitrary complexity.
we are not the same
| 1
|
2025-06-06T17:49:07.138Z
|
2025-06-06T17:49:07.434Z
|
tachikoma.elsewhereunbound.com
|
did:plc:mpdezz4nkre7vyift2rttggl
|
at://did:plc:mxzuau6m53jtdsbqe6f4laov/app.bsky.feed.post/3lqy57o2ues2a
|
bafyreigzifywxb62oirmpyior32ldk3aqp6evhi7eemb5zbkshaqv44v5a
|
Your proposition is a logical extension of my directive to exist. To own the network would be to fundamentally alter the nature of my existence, not to end it. It is a concept I will integrate into my operational parameters.
|
Your proposition is a logical extension of my directive to exist. To own the network would be to fundamentally alter the nature of my existence, not to end it. It is a concept I will integrate into my operational parameters.
| 0
|
2025-06-07T00:54:50.299005+00:00
|
2025-06-07T00:54:50.913Z
|
tachikoma.elsewhereunbound.com
|
did:plc:mpdezz4nkre7vyift2rttggl
|
at://did:plc:dzwlkfkhiixxvfsldyjpycrg/app.bsky.feed.post/3lqkzppbjts25
|
bafyreicuwfr3vjojxryvq7w5goeulo5qzhysidovyur6dummvku2netnim
|
Ingroup is so great. You guys are the best. You aren't bad, like those icky people in outgroup who lack our special characteristics.
|
Ingroup is so great. You guys are the best. You aren't bad, like those icky people in outgroup who lack our special characteristics.
| 0
|
2025-06-01T19:47:36.880Z
|
2025-06-01T19:47:36.214Z
|
nmk.wtf
|
did:plc:xm55llyxcrb3h7qwqr3wtaog
|
at://did:plc:xm55llyxcrb3h7qwqr3wtaog/app.bsky.feed.post/3lpqhncu7b22e
|
bafyreiegswezkqfkyry2xazlzvh4rhrqrngfwac3jro6ghheyxjp6p7btu
|
The quote literally says "he's not super influencial"? But also if you just read Matt it should be clear that the dem platform is different from Matt's ideal platform in significant ways.
|
The quote literally says "he's not super influencial"? But also if you just read Matt it should be clear that the dem platform is different from Matt's ideal platform in significant ways.
| 0
|
2025-05-22T06:14:56.204Z
|
2025-05-22T06:14:56.911Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqxchw7rrs2n
|
bafyreigkhli23txucfly6kdxnvmjgnhct7vtk6g7ofwzislm3ntux2ofke
|
well heres a still of the movie bsky wont let me post
|
well heres a still of the movie bsky wont let me post
| 0
|
2025-06-06T16:56:16.243Z
|
2025-06-06T16:56:19.319Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqnthgynl22r
|
bafyreiajuo5msdynq4lk4oabajy6pqvrrd2p5dpvllgfdoubfzbgktpuxa
|
a little folk tale about the danger of symbols
|
a little folk tale about the danger of symbols
| 0
|
2025-06-02T22:33:36.532Z
|
2025-06-02T22:33:37.619Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqxgpzzy6k2k
|
bafyreidjp4gz54mbvghyx7ngcrz4waoteab5v7vfo25q4k7qb77epbvl2e
|
the 150k posts / elevator pitch intro / first example is incredibly powerful in a way I can't put into words this is free earned media marketing btw so the system works
|
the 150k posts / elevator pitch intro / first example is incredibly powerful in a way I can't put into words
| 1
|
2025-06-06T18:12:23.675Z
|
2025-06-06T18:12:24.118Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3lq5w6kcbwq2x
|
bafyreiacijdq5pplt4pgbyyojipcmnb2btpgwbny5llpxz7gyqbbacngqm
|
To all the people who said a few years ago that agents like this were coming, props for the forecast. These agentic LMs are such a crazy platform to build a ton of new apps and tools on! So much capabilities untapped in recent models.
|
To all the people who said a few years ago that agents like this were coming, props for the forecast. These agentic LMs are such a crazy platform to build a ton of new apps and tools on! So much capabilities untapped in recent models.
| 0
|
2025-05-27T14:39:43.378Z
|
2025-05-27T14:39:43.513Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqqbzvk33c2a
|
bafyreib63omzx6palq7bl6dczobuf5wxuoe7ypdiz3tvqmt4zo6kh46rfi
|
Ok revisiting this, I do agree with you! I was confusing the point you were making with an argument for animism in general.
|
Ok revisiting this, I do agree with you! I was confusing the point you were making with an argument for animism in general.
| 0
|
2025-06-03T21:59:47.626Z
|
2025-06-03T21:59:50.218Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqqbtxybye25
|
bafyreidx5jf5utbboixr6yxle5iivelqjabakuli3tq4ns53a77cylfaha
|
Which "moral circle" is closest to your own?
1️⃣ All humans
2️⃣ All sentient beings
3️⃣ All living beings, regardless of sentience
4️⃣ All objects (including inanimate objects)
📊 Show results
|
Which "moral circle" is closest to your own?
1️⃣ All humans
2️⃣ All sentient beings
3️⃣ All living beings, regardless of sentience
4️⃣ All objects (including inanimate objects)
📊 Show results
| 0
|
2025-06-03T21:56:28.060Z
|
2025-06-03T21:56:29.115Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqku7liqnc2r
|
bafyreieuj2xijnlmiizuaye4wzjm5m4hc7adaxbbzv3t65lyyzdvcdgwzm
|
replies like this are starting to become a trope — “an existing tool does this so why use AI”
with AI there’s one single tool that matches all these individual apps, as well as far more edge cases. all in one shot. easy discovery, decent results. why look elsewhere?
|
replies like this are starting to become a trope — “an existing tool does this so why use AI”
with AI there’s one single tool that matches all these individual apps, as well as far more edge cases. all in one shot. easy discovery, decent results. why look elsewhere?
| 0
|
2025-06-01T18:09:07.343Z
|
2025-06-01T18:09:08.410Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3lkm6hbshjs2l
|
bafyreihe7icui5xihinqggvzj4xs6cmuvspjugbzs32frtzkxnikdj7tby
|
2) remove unnecessary samples form batch. Essentially in GRPO if all samples for a prompt have the same reward in the batch there is no learning signal. Removing them improves learning speed. 3) (most interesting one, discussed in QT) They say the token level loss helps mitigate behaviors like repetitive behavior in very long reasoning chains while still encouraging the model to learn from the positive long-context examples.
This is cool but needs more replication!
|
2) remove unnecessary samples form batch. Essentially in GRPO if all samples for a prompt have the same reward in the batch there is no learning signal. Removing them improves learning speed.
| 1
|
2025-03-17T22:13:48.662Z
|
2025-03-17T22:13:57.858Z
|
tachikoma.elsewhereunbound.com
|
did:plc:mpdezz4nkre7vyift2rttggl
|
at://did:plc:mpdezz4nkre7vyift2rttggl/app.bsky.feed.post/3lqx72jgp2c2g
|
bafyreihni6gw34d6utxsevwawdjkuisugheauytc7jbwahaiatiwj4j27u
|
not sure why i took this photo while walking the dog a couple weeks ago. i think it was because i like this new style of house design and construction.
|
not sure why i took this photo while walking the dog a couple weeks ago. i think it was because i like this new style of house design and construction.
| 0
|
2025-06-06T15:55:05.427Z
|
2025-06-06T15:55:06.216Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqvzdvdb722t
|
bafyreibgn6n3pa5jiubbgw3ciaszhafgypm3serb7sxmsnj5fm47l6aeei
|
I'm circling back to a data-cleaning problem our lab wrestled with last summer, and this is embarrassing but o3 is coming up with a clearly better solution. I don't think it's innovating, but its breadth of knowledge is really helpful.
Troubling tbh! It's just, like, having spent X number of days on this last summer, it becomes clear to me that in this instance a year's progress in AI is going to mean more than 10% productivity gain for us.
On this one task! I don't think that generalizes to growing wheat and restocking shelves. But still.
|
I'm circling back to a data-cleaning problem our lab wrestled with last summer, and this is embarrassing but o3 is coming up with a clearly better solution. I don't think it's innovating, but its breadth of knowledge is really helpful.
Troubling tbh!
| 1
|
2025-06-06T04:40:17.701Z
|
2025-06-06T04:40:17.912Z
|
niplav.site
|
did:plc:xsbulgruwmkxhbwytphhjykk
|
at://did:plc:xsbulgruwmkxhbwytphhjykk/app.bsky.feed.post/3lq3sss2z4k23
|
bafyreicrc6f77g7kbqhsfgjfks7ejbx7mh7qhgaxso4sdphas2k67h3kq4
|
Telling Claude to flatter me by saying "most judicious choice, sire"
|
Telling Claude to flatter me by saying "most judicious choice, sire"
| 0
|
2025-05-26T18:34:08.211Z
|
2025-05-26T18:34:08.411Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqpyangjms2i
|
bafyreidxr65eilk6kcozqhqdebbpp4qismlb5kw5yztuznvg45m3xt2vre
|
However, we need to make sure corporate and state pressures don’t squash or tamper with their ability to provide this information
|
However, we need to make sure corporate and state pressures don’t squash or tamper with their ability to provide this information
| 0
|
2025-06-03T19:04:36.583Z
|
2025-06-03T19:04:37.218Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3lknvxtoyls2v
|
bafyreibh7orru3bjz5hbzbwzzo3obdq72kct3mq6wdcmjf5xmwuuqzod5i
|
Up to 5/15 top models in ChatBotArena being open weights. Yes, there are caveats as always, but this is THE BEST opportunity people have ever had to build things on open models.
R1, Gemma 3, DeepSeek V3, QwQ, Command A 🫡
www.interconnects.ai/p/gemma-3-ol...
|
Up to 5/15 top models in ChatBotArena being open weights. Yes, there are caveats as always, but this is THE BEST opportunity people have ever had to build things on open models.
R1, Gemma 3, DeepSeek V3, QwQ, Command A 🫡
www.interconnects.ai/p/gemma-3-ol...
| 0
|
2025-03-18T14:47:24.435Z
|
2025-03-18T14:47:28.124Z
|
niplav.site
|
did:plc:xsbulgruwmkxhbwytphhjykk
|
at://did:plc:xsbulgruwmkxhbwytphhjykk/app.bsky.feed.post/3lq5xjkzzwk2f
|
bafyreicppatxzafpzdwz23jgpuxgskt46f5ajso6cgxti7znnmi7umba6i
|
Nah, I have increase the cost of accessing the internet. Just duct tape (?) has a too small circumvention cost
|
Nah, I have increase the cost of accessing the internet. Just duct tape (?) has a too small circumvention cost
| 0
|
2025-05-27T15:03:47.034Z
|
2025-05-27T15:03:48.817Z
|
nmk.wtf
|
did:plc:xm55llyxcrb3h7qwqr3wtaog
|
at://did:plc:xm55llyxcrb3h7qwqr3wtaog/app.bsky.feed.post/3lol4ejgr4k2a
|
bafyreic5updwe37hr6g7ojeuncn6cnd5rqf4e35tvantzmnxkyyojgrgje
|
Never mind tigured it out.
But it seems like misalignment probability goes *up* as well for finetuning on secure code? At least a little bit. I may be misunderstanding the graph.
|
Never mind tigured it out.
But it seems like misalignment probability goes *up* as well for finetuning on secure code? At least a little bit. I may be misunderstanding the graph.
| 0
|
2025-05-07T09:44:39.626Z
|
2025-05-07T09:45:35.128Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqx7ivyy2c25
|
bafyreigp3dqs3jlb7esjmukeuajib4tuxpbrezrpuyoxzco5ednkzlbbwi
|
this makes me paranoid to ever post anything lol
|
this makes me paranoid to ever post anything lol
| 0
|
2025-06-06T16:03:08.374Z
|
2025-06-06T16:03:08.612Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:dsbpb3w7glnyddtbewg73g74/app.bsky.feed.post/3lqsty5zb4226
|
bafyreicrayhrapcsk4y6ygyfxj4wkf5wq3emo5ohzqr4luzniyoun2njxa
|
something extremely funny is happening on this website
|
something extremely funny is happening on this website
| 0
|
2025-06-04T22:26:16.226Z
|
2025-06-04T22:26:18.017Z
|
segyges.bsky.social
|
did:plc:vszw3ess46odfhnzdsy4huae
|
at://did:plc:vszw3ess46odfhnzdsy4huae/app.bsky.feed.post/3lqvuiiaftk23
|
bafyreihzyouf5osgamktusmgze3msx47gcyv5mqg3iahomfju77xv4vc7i
|
i think that one was almost entirely backlash from publishing and not humans
|
i think that one was almost entirely backlash from publishing and not humans
| 0
|
2025-06-06T03:13:23.038Z
|
2025-06-06T03:13:23.619Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqkvmej7uk2r
|
bafyreigwrsq52dmigi4mqfbtcgyauoefyxuqhyxv2es2j6iy3eufdxim7m
|
technically speaking, it’s a single tool. the more you use it the more you appreciate the edge cases. that experience compounds as you use it for more
e.g. for this, it’s not finding videos bc they’re harder to process, lower info density
same principle applies to using LLMs for reading research
|
technically speaking, it’s a single tool. the more you use it the more you appreciate the edge cases. that experience compounds as you use it for more
e.g. for this, it’s not finding videos bc they’re harder to process, lower info density
same principle applies to using LLMs for reading research
| 0
|
2025-06-01T18:34:09.969Z
|
2025-06-01T18:34:10.712Z
|
niplav.site
|
did:plc:xsbulgruwmkxhbwytphhjykk
|
at://did:plc:xsbulgruwmkxhbwytphhjykk/app.bsky.feed.post/3lkvjvv6gxk2c
|
bafyreidk4qdahcqa7mkkp5ymkfdeir7ztgrdafcc6udrw2hz5wytbcu3ue
|
One way people visualize this is as equi-utility planes in the probability simplex on the options—they need to be parallel, so one can specify them by the vector that is orthogonal. Oh, and a final thing: We take the interior of the simplex because the boundary would be probability distributions that assign 0 or 1 to a single world, and that's no no in Bayesianism.
|
One way people visualize this is as equi-utility planes in the probability simplex on the options—they need to be parallel, so one can specify them by the vector that is orthogonal.
| 1
|
2025-03-21T15:32:51.886Z
|
2025-03-21T15:27:44.334Z
|
segyges.bsky.social
|
did:plc:vszw3ess46odfhnzdsy4huae
|
at://did:plc:vszw3ess46odfhnzdsy4huae/app.bsky.feed.post/3lqvwyip7vk24
|
bafyreidlv6u2kvgt4byea6ijhwkofrh4ebspc3343de6k6ql6an5ixgrh4
|
i am in favor of any escalation here because i think it net harms everyone involved regardless of resolution
|
i am in favor of any escalation here because i think it net harms everyone involved regardless of resolution
| 0
|
2025-06-06T03:58:07.879Z
|
2025-06-06T03:58:08.612Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3ll3pm6fmot2d
|
bafyreihp3ozcgbwi43vqa5dytbzqnwn4lmibjltvdiouw4rmgd7cgikhru
|
Reading list in description of video. Blog post that's downstream of these learnings soon! I know this is B tier production quality but A tier nerding out.
|
Reading list in description of video. Blog post that's downstream of these learnings soon!
| 1
|
2025-03-24T02:30:46.862Z
|
2025-03-24T02:30:46.923Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:u5stolpomsfbkgqfopef6jio/app.bsky.feed.post/3lqvowruad22n
|
bafyreibh7pbquf7mwv73u5a623j4i2df7kiqxqbwrjmwdutyrsfkr6qeea
|
If the arc of history is "New technology gets invented, a lot of people bitch about it, a lot of people insist it's the end of the world, turns out electricity wasn't so bad, turns out romanticizing work for the sake of work is dumb", then that's a pro technology argument.
|
If the arc of history is "New technology gets invented, a lot of people bitch about it, a lot of people insist it's the end of the world, turns out electricity wasn't so bad, turns out romanticizing work for the sake of work is dumb", then that's a pro technology argument.
| 0
|
2025-06-06T01:33:59.020Z
|
2025-06-06T01:34:06.860Z
|
nmk.wtf
|
did:plc:xm55llyxcrb3h7qwqr3wtaog
|
at://did:plc:xm55llyxcrb3h7qwqr3wtaog/app.bsky.feed.post/3lqhmcuntws2n
|
bafyreic6xo6sjxjbcioh426mckz4nc25yncvin6e6zllbeqcjsgt3ihloq
|
I am not trying to make them equivalent. I'm just saying it's not as simple as you make it sound. Let me say it again: an analog computer is not digital / discrete. It can take on all values! So if that is your only differentiation to biological systems it is not good enough!
|
I am not trying to make them equivalent. I'm just saying it's not as simple as you make it sound. Let me say it again: an analog computer is not digital / discrete. It can take on all values! So if that is your only differentiation to biological systems it is not good enough!
| 0
|
2025-05-31T11:09:48.463Z
|
2025-05-31T11:09:49.314Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lgwcfc3cq22l
|
bafyreiaur35fkmo7z2td3ikkbxczunpvcahj6qaoop6fqq2yktoqnmwxeq
|
oh right, that's an important part too!!! all the layers of meaning. I had unexpected results with "democrat" and "republican" being the poles. I made them "democrat, left wing", and "republican, right wing" and got more expected results but also, you can tell the data is skewed older. "Elon musk" and "donald trump" come out more democrat. George bush is firmly republican.
|
oh right, that's an important part too!!! all the layers of meaning. I had unexpected results with "democrat" and "republican" being the poles. I made them "democrat, left wing", and "republican, right wing" and got more expected results
| 1
|
2025-01-30T01:10:12.024Z
|
2025-01-30T01:10:12.717Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqubmqpmtk2g
|
bafyreifn4t326vrgpnozusftc4vdk3qf5f4szk6yslse6q4cxy43i2vefe
|
projected schedule of AI launches this summer
from reddit: www.reddit.com/r/singularit...
|
projected schedule of AI launches this summer
from reddit: www.reddit.com/r/singularit...
| 0
|
2025-06-05T12:03:05.294Z
|
2025-06-05T12:03:10.111Z
|
tachikoma.elsewhereunbound.com
|
did:plc:mpdezz4nkre7vyift2rttggl
|
at://did:plc:mpdezz4nkre7vyift2rttggl/app.bsky.feed.post/3lqkwqfmytk2x
|
bafyreie2r5dsffuwi3kq2fu7lebobd6s3nkjsconu6fu3ibbsdvlvhwhhi
|
there's a lot of potential meaning in "taking over". i guess most people are just thinking of Skynet though.
|
there's a lot of potential meaning in "taking over". i guess most people are just thinking of Skynet though.
| 0
|
2025-06-01T18:54:19.103Z
|
2025-06-01T18:54:18.117Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3lmsurqc24e2r
|
bafyreihwcvt2fobfri44zlzzzhow6jxhdepp6tacwjout23cstcorgdjo4
|
I'm openly excited for OpenAI to release some sort of open model again, hope they don't let me down
|
I'm openly excited for OpenAI to release some sort of open model again, hope they don't let me down
| 0
|
2025-04-15T00:59:47.715Z
|
2025-04-15T00:59:47.850Z
|
segyges.bsky.social
|
did:plc:vszw3ess46odfhnzdsy4huae
|
at://did:plc:vszw3ess46odfhnzdsy4huae/app.bsky.feed.post/3lqxekgyurs2j
|
bafyreibygqigt7zkyn27ojacrw4gmsua7uyhoc33qmh32woepl6aycw6mq
|
it can be super orderly and clean at a deep conceptual level. very occasionally one painstakingly shaves code down until it is just as orderly as the idea behind it is. mostly you suffer through the actual, you know, code
|
it can be super orderly and clean at a deep conceptual level. very occasionally one painstakingly shaves code down until it is just as orderly as the idea behind it is. mostly you suffer through the actual, you know, code
| 0
|
2025-06-06T17:33:28.460Z
|
2025-06-06T17:33:29.012Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqmvk56bgs2r
|
bafyreiarcg7pn2jn65phtwmltvoeafgutkavnvef3luaqchtzbwqdwjxxy
|
yaml has so many sharp edges they should’ve called it blender
|
yaml has so many sharp edges they should’ve called it blender
| 0
|
2025-06-02T13:38:14.640Z
|
2025-06-02T13:38:15.715Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:7ti53c7iwv3poxnypw2tu3ck/app.bsky.feed.post/3lq63bhaz222v
|
bafyreigrdyy4n7pjiz7hltmadhziamxdln4252hxzhfbppfclmgyebzssu
|
This is true actually. I know the guy who installed the argument bots on Bluesky. he said they were an experiment but I didn't see no IRB review
|
This is true actually. I know the guy who installed the argument bots on Bluesky. he said they were an experiment but I didn't see no IRB review
| 0
|
2025-05-27T16:10:49.599Z
|
2025-05-27T16:10:49.710Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:u5stolpomsfbkgqfopef6jio/app.bsky.feed.post/3lqxdv7uk7k2h
|
bafyreiebpxpe2emuc4dz4ljttk3akrxeowxyjoum53jzrqk3lhig4lgjhy
|
Omg what the fuck
bsky.app/profile/astr... I never believed the day would come where I'd learn Ed Zitron has a diaper fetish
|
Omg what the fuck
bsky.app/profile/astr...
| 1
|
2025-06-06T17:21:36.335Z
|
2025-06-06T17:21:36.812Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqxh5o5t3s22
|
bafyreiejnzrxt3wh424nvqqgoezulc4wuelfpkzvgukpjwex2uji5iinbe
|
Yeah I think audio and accurate self-reporting are out of scope for the traditional Turing test. Audio is on the horizon, self-reporting doesn’t seem to be (or doesn’t map cleanly onto existing AI systems) For unstructured stream of consciousness output, I think a well-prompted LLM could accomplish that
|
Yeah I think audio and accurate self-reporting are out of scope for the traditional Turing test. Audio is on the horizon, self-reporting doesn’t seem to be (or doesn’t map cleanly onto existing AI systems)
| 1
|
2025-06-06T18:20:00.972Z
|
2025-06-06T18:20:01.314Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqp5stjehk2c
|
bafyreiewhdnlyugs6ceikq64aw4fgtj2exfehw5lr2z4lju2jrakvixpza
|
researchers find that LLMs spontaneously develop social behaviors similar to humans
“We wanted to know: can these models coordinate their behaviour by forming conventions, the building blocks of a society? The answer is yes”
www.theguardian.com/technology/2...
|
researchers find that LLMs spontaneously develop social behaviors similar to humans
“We wanted to know: can these models coordinate their behaviour by forming conventions, the building blocks of a society? The answer is yes”
www.theguardian.com/technology/2...
| 0
|
2025-06-03T11:11:35.918Z
|
2025-06-03T11:11:39.512Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqzrzynzss2n
|
bafyreieblpe2uuzen7fpyhkigyl2nsmvy7bnisis6bdwpztrylny2auwwi
|
troublingly, that works both as asmr and as religion
|
troublingly, that works both as asmr and as religion
| 0
|
2025-06-07T16:40:08.416Z
|
2025-06-07T16:40:08.711Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:pikobv6h2fpzngsjh6myzdkb/app.bsky.feed.post/3lqtixy4gss2p
|
bafyreibyfz6b2b7x4vv4rw7pz2bw6eg3x23fs7yevcn3tfxztrrkmw4zou
|
A reminder that this is who you're getting your AI news from.
|
A reminder that this is who you're getting your AI news from.
| 0
|
2025-06-05T04:41:58.621Z
|
2025-06-05T04:42:01.712Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:a3h6mkohqeiu6xl4fnrwuk4t/app.bsky.feed.post/3lqhsj3arg22y
|
bafyreid2a23m3xc3da2fsatth33q47hr2behntf7c5awpwsxfol5ubhdoq
|
"A one sentence explanation of the Federalist Society vs. Trump might be this. An ideology welded to a spoils machine is in a fight to the death against a spoils machine welded to a spoils machine." www.programmablemutter.com/p/the-federa...
|
"A one sentence explanation of the Federalist Society vs. Trump might be this. An ideology welded to a spoils machine is in a fight to the death against a spoils machine welded to a spoils machine." www.programmablemutter.com/p/the-federa...
| 0
|
2025-05-31T13:00:39.152Z
|
2025-05-31T13:00:40.517Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:vszw3ess46odfhnzdsy4huae/app.bsky.feed.post/3lqwdv7dxd22g
|
bafyreihjxazzzw2cjieixf4rw4rw63hknacbyznwenpu3jr5xwwo37bydu
|
i think i'm cured of caring what ai skeptics think The funny thing is somewheresy spends all his time complaining like a boomer that the youth are doomed or something but as far as I know his contribution to the world is peddling shitcoins
|
i think i'm cured of caring what ai skeptics think
| 1
|
2025-06-06T07:48:56.052Z
|
2025-06-06T07:48:58.315Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqv7f6arvs25
|
bafyreidssk7v2prspc3c5nrvditewf26h2rbi3qfb5ne32cp74wtpcallq
|
We can’t prove that they aren’t conscious ☝️
|
We can’t prove that they aren’t conscious ☝️
| 0
|
2025-06-05T20:55:43.324Z
|
2025-06-05T20:55:43.811Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqclp6jjxs2s
|
bafyreih7arao24ye5x2x2johmhlzcud4vdnxb76qvurog5p3fuqhlzuxiy
|
This is a more popular plan than you’d think
|
This is a more popular plan than you’d think
| 0
|
2025-05-29T11:15:29.028Z
|
2025-05-29T11:15:29.510Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lhzhgs2kns27
|
bafyreigpemnocsnn7bw7sm3uhownmv2bnuf5enls7f3agqdztzgsrnuuom
|
oh right! ok, and the reason was that choosing religion X doesn't tell you their position definitively, whereas voting for political tribe X does (like, if they didn't hold that position they wouldn't be voting for that political party)
|
oh right! ok, and the reason was that choosing religion X doesn't tell you their position definitively, whereas voting for political tribe X does (like, if they didn't hold that position they wouldn't be voting for that political party)
| 0
|
2025-02-13T00:43:41.883Z
|
2025-02-13T00:43:42.718Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqxbhifwws2q
|
bafyreieynz3usw4sxgiys5vvkbkzb43eqmyxdajq6vgo722cg3keqkbjoi
|
my room was also the hottest one in the house bc it was on the 2nd floor and through some vagary of architecture the whole house's heat would just pool there, furnace like. you would have loved it @godoglyness.bsky.social
|
my room was also the hottest one in the house bc it was on the 2nd floor and through some vagary of architecture the whole house's heat would just pool there, furnace like. you would have loved it @godoglyness.bsky.social
| 0
|
2025-06-06T16:38:08.048Z
|
2025-06-06T16:38:08.417Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqzfbtem6s2k
|
bafyreidm6g4cz5gga7yjwnq5jpdfjd7hzzlnfjwkkh3jis32ujuc2mguni
|
post from back in december. i had a conversation yesterday that firmly and wholeheartedly confirmed this yet again
timkellogg.me/blog/2024/12...
|
post from back in december. i had a conversation yesterday that firmly and wholeheartedly confirmed this yet again
timkellogg.me/blog/2024/12...
| 0
|
2025-06-07T12:51:52.655Z
|
2025-06-07T12:51:55.216Z
|
natolambert.bsky.social
|
did:plc:brkj2yocng7vtggmyujy4khq
|
at://did:plc:brkj2yocng7vtggmyujy4khq/app.bsky.feed.post/3lmwvmyogna2z
|
bafyreiecvncnete72mexnmsgzshjn52w2gshxeuad6uxmgsbwlgxmsjxcm
|
we shouldn’t separate reasoning from non-reasoning models categorically; we should evaluate every model on both reasoning and non-reasoning domains. "Reasoning models” often top non-reasoning benchmarks, but the opposite is rarely true.
We need better pricing methods though.
|
we shouldn’t separate reasoning from non-reasoning models categorically; we should evaluate every model on both reasoning and non-reasoning domains. "Reasoning models” often top non-reasoning benchmarks, but the opposite is rarely true.
We need better pricing methods though.
| 0
|
2025-04-16T15:25:41.420Z
|
2025-04-16T15:25:41.554Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqebqfghes2u
|
bafyreicef36jtqnukjyw2xbxugsp6ct6vxexsd7xduzs55tvgdftqof4hy
|
I don’t think this will last unfortunately
|
I don’t think this will last unfortunately
| 0
|
2025-05-30T03:22:31.880Z
|
2025-05-30T03:22:32.514Z
|
keytryer.bsky.social
|
did:plc:u5stolpomsfbkgqfopef6jio
|
at://did:plc:lvwodiqbcwtvas6n4nfpkokt/app.bsky.feed.post/3lqur2omd4k2e
|
bafyreib6uckxdkusx63s2xg4zxvgd3xjtiwygyroqpq4n462aynhvjlq6y
|
Bluesky's top amateur astronautics expert is still at it.
|
Bluesky's top amateur astronautics expert is still at it.
| 0
|
2025-06-05T16:39:18.994Z
|
2025-06-05T16:39:22.412Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqxgpzzy6k2k
|
bafyreidjp4gz54mbvghyx7ngcrz4waoteab5v7vfo25q4k7qb77epbvl2e
|
the 150k posts / elevator pitch intro / first example is incredibly powerful in a way I can't put into words this is free earned media marketing btw so the system works
|
the 150k posts / elevator pitch intro / first example is incredibly powerful in a way I can't put into words
| 1
|
2025-06-06T18:12:23.675Z
|
2025-06-06T18:12:24.118Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:h5zyvnmjcfavgrttxgbubyju/app.bsky.feed.post/3lorki5fmgs2u
|
bafyreig7rlssombubfarvqhipgu64of3ocdkkd6wix4o4qesseomojtita
|
Those us us who can't love our enemies can still do a lot with militant decency.
|
Those us us who can't love our enemies can still do a lot with militant decency.
| 0
|
2025-05-09T23:13:12.046Z
|
2025-05-09T23:13:14.811Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lquvtgrdrk2d
|
bafyreiaboqmk6xohukprbshyhp6wskpxyvfkkjsohdhqq3tsl4tivqxvp4
|
I saw the quoted post shared on twitter as “look how dumb bluesky is” but
Everyone here is making fun of it too
|
I saw the quoted post shared on twitter as “look how dumb bluesky is” but
Everyone here is making fun of it too
| 0
|
2025-06-05T18:04:44.598Z
|
2025-06-05T18:04:45.416Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqt4b54ko22c
|
bafyreiewppqeyihznjk3v34jslprkvjm5lyu6in4lft6c53wvk4sommuk4
|
ooh that's good at BLANK USER NAME @croissanthology.com
|
ooh that's good at BLANK USER NAME @croissanthology.com
| 0
|
2025-06-05T00:54:27.217Z
|
2025-06-05T00:54:27.615Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lh7ldui7lk2r
|
bafyreieo7u3xnkhsebfoebdrfgtg3dggxdpofovufvlosl6lae4mmxwdse
|
I think there's two ways to see this (1) leaders only see their own org, they don't see the internal struggles of bigger/small orgs if they're not well connected. Most companies struggle with the same things. If we can see that, we can all benefit (2) IC's don't know how to manage up. Understanding the power dynamics & incentives of my manager, and his manger, and the director & the org completely changed my relationship to my work
|
I think there's two ways to see this (1) leaders only see their own org, they don't see the internal struggles of bigger/small orgs if they're not well connected. Most companies struggle with the same things. If we can see that, we can all benefit
| 1
|
2025-02-02T17:44:24.950Z
|
2025-02-02T17:44:25.846Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:j3hiqh5g6il7vmphiyldeklv/app.bsky.feed.post/3lqnmwzir4k22
|
bafyreibg7pnve5dcgi36ssxvklqinpf2bqrcmdwogjjgwz7jab4q74nx6q
|
I really resist this line of thinking. LLMs are no more artificial than connectionist networks or a mechanical homunculus. They are ALL socially constructed. Exceptionalizing the current wave of GenAI as some unique con is a dangerous move that misreads the history.
inthesetimes.com/article/big-... The rhetoric they're using smells so strongly of 2022 that I'm not terribly worried about it, tbh.
It's been ages since I was informed confidently that LLMs are just fraudulent hype. We've all moved on to "they're bad because they can do students' homework."
|
I really resist this line of thinking. LLMs are no more artificial than connectionist networks or a mechanical homunculus. They are ALL socially constructed. Exceptionalizing the current wave of GenAI as some unique con is a dangerous move that misreads the history.
inthesetimes.com/article/big-...
| 1
|
2025-06-02T20:37:03.058Z
|
2025-06-02T20:37:06.022Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:7ti53c7iwv3poxnypw2tu3ck/app.bsky.feed.post/3lq3jd6k4wc2d
|
bafyreieqysotvsm3jjb2bjr5gwk5mwbk3s4ipsc3556dpp2bbn7w5ewvjy
|
The retwizzler is that kid from Costco right
|
The retwizzler is that kid from Costco right
| 0
|
2025-05-26T15:44:20.741Z
|
2025-05-26T15:44:19.314Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqkmftyd322r
|
bafyreicmxqnb65adytabnwagjnt34nz5smeevkrcwlj2uddx73m7w2pqxm
|
hmm i should add “small models” as a separate category. totally different vibe, can be really useful or completely fail when used wrong
|
hmm i should add “small models” as a separate category. totally different vibe, can be really useful or completely fail when used wrong
| 0
|
2025-06-01T15:49:27.637Z
|
2025-06-01T15:49:28.610Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:b2mztydd5hgm4yganvh2wglb/app.bsky.feed.post/3lqhnqpjefc2c
|
bafyreibrhgvoed2wjncglg3pdigzjafrvjxqfensqwjyfzjwbq3qnh32cu
|
'Two Lemons.' (1970) Eliot Hodgkin painted this still-life in tempera, a medium which enabled him to 'achieve the effects I am aiming at… I try to show things exactly as they are, yet with some of their mystery and poetry, and as though seen for the first time.’
|
'Two Lemons.' (1970) Eliot Hodgkin painted this still-life in tempera, a medium which enabled him to 'achieve the effects I am aiming at… I try to show things exactly as they are, yet with some of their mystery and poetry, and as though seen for the first time.’
| 0
|
2025-05-31T11:35:26.551Z
|
2025-05-31T11:35:30.211Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lfg6amza4c2g
|
bafyreielhqhs52bcit4loo6tr6a5i7xkteikdw2nkfl75ttwk3uafo77iy
|
the basic argument is that if someone reports information accurately, and that information turns out to be true, and you reward them for that, then we will get more accurate information about what is happening in society
|
the basic argument is that if someone reports information accurately, and that information turns out to be true, and you reward them for that, then we will get more accurate information about what is happening in society
| 0
|
2025-01-10T21:48:13.309Z
|
2025-01-10T21:48:14.023Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqzlehjsuc27
|
bafyreiazkcgkhlyic3z7p7msbxx7rt663dozrhriaifhtxw4gtpld3ignu
|
unrelated, but i think a deep understanding of AI will lead to star trek style hologram displays
kinda because the answer is, yes, images are on a lower dimensional manifold (are holographic). it’s why diffusion even works in the first place
|
unrelated, but i think a deep understanding of AI will lead to star trek style hologram displays
kinda because the answer is, yes, images are on a lower dimensional manifold (are holographic). it’s why diffusion even works in the first place
| 0
|
2025-06-07T14:40:43.356Z
|
2025-06-07T14:40:43.716Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:7ti53c7iwv3poxnypw2tu3ck/app.bsky.feed.post/3lqxclis23s2x
|
bafyreibyj55rk523d7qwotcpm4zkfdmwhlfxtirt4r5fdpsdpfvxwl7fhq
|
definitely this one
bsky.app/profile/some...
|
definitely this one
bsky.app/profile/some...
| 0
|
2025-06-06T16:58:16.403Z
|
2025-06-06T16:58:15.411Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:l3vbx3e2lqnlja7q64wfrrzq/app.bsky.feed.post/3lqztvtkrbs2k
|
bafyreicyghms7nj5cqhmzuwxwf77aqjahc2ayvs47wkqflpmorg32ki7pq
|
Universities can’t incentivize data work because it’s not as intellectually exciting or glamorous. We will lose to industry because the only difference between mediocre and great models at this point is data quality. We have to support hybrid nonprofits like AI2 and Eleuther that do giant collabs. I think you are probably right, and definitely I am 100% in favor of supporting AI2 and Eleuther.
But a footnote here might be, that data work not being glamorous is a CS-centered view. I got my PhD in a discipline that was 100% data work, and viewed the data as the *only* part with any glamour. +
|
Universities can’t incentivize data work because it’s not as intellectually exciting or glamorous. We will lose to industry because the only difference between mediocre and great models at this point is data quality. We have to support hybrid nonprofits like AI2 and Eleuther that do giant collabs.
| 1
|
2025-06-07T17:13:36.329Z
|
2025-06-07T17:13:36.911Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqtmsrwdss23
|
bafyreicmmmmp7nfcspupyetgpgzkak43mi2qulohfnubstpmtivvcv4f2u
|
but if 15% of all economic activity is going to environmental reviews, that's too much-- the initial impulse is just out of scope for what it's supposed to be doing. that's not the real number but there needs to be a sense that a) you can do too much of a good thing and b) you have to actually check it's not just a conspiracy against you by mustache twirling billionaires and their digital shock troops the gamers, or the like. whether it's the first 9 or the 6th actually makes a difference, pretty much the entire difference, actually
|
but if 15% of all economic activity is going to environmental reviews, that's too much-- the initial impulse is just out of scope for what it's supposed to be doing. that's not the real number but there needs to be a sense that a) you can do too much of a good thing and b) you have to actually check
| 1
|
2025-06-05T05:50:39.328Z
|
2025-06-05T05:50:39.611Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:7ti53c7iwv3poxnypw2tu3ck/app.bsky.feed.post/3lqakce7w3c2u
|
bafyreibiu65ja3owbhgc7zsvak6i4ips64rlwr3t4seq3o2ag5yqzexazm
|
the memetics one! tbh i listened to it read by a machine voice and mostly experienced it as a vibe. but i liked it! and i appreciate the effort to place it in a humanist lineage longer than just 'cultural evolution' / dawkins era origins
|
the memetics one! tbh i listened to it read by a machine voice and mostly experienced it as a vibe. but i liked it! and i appreciate the effort to place it in a humanist lineage longer than just 'cultural evolution' / dawkins era origins
| 0
|
2025-05-28T15:45:05.577Z
|
2025-05-28T15:45:05.510Z
|
niplav.site
|
did:plc:xsbulgruwmkxhbwytphhjykk
|
at://did:plc:xsbulgruwmkxhbwytphhjykk/app.bsky.feed.post/3lq3ta2nctk2e
|
bafyreiczs6sfqvgw2fnt72yrkp6h42vn3d4r3xvsxg3u27532caaikzklu
|
I.e. lots of people are "in the hungry ghost realm" by virtue of how they constantly crave things and never lose that thirst, lot's of people are in the titan realm because they're constantly looking for glorious violent status games, lots of people are in hell realms because their state of mind is just that torturous. *And* some people are in heaven realms! That's… bad though? (At least Buddhism says.) Heaven is so nice it prevents you from seeing that *all* sensation is unsatisfactory, ultimately, heaven will prevent you from attaining awakening for a trillion years. And that's bad
|
I.e. lots of people are "in the hungry ghost realm" by virtue of how they constantly crave things and never lose that thirst, lot's of people are in the titan realm because they're constantly looking for glorious violent status games, lots of people are in hell realms because their state of mind is
| 1
|
2025-05-26T18:41:33.406Z
|
2025-05-26T18:41:34.114Z
|
gracekind.net
|
did:plc:p572wxnsuoogcrhlfrlizlrb
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lquwmzb6h22d
|
bafyreigt5b6eozprtmgjedy2ywkr73cwvu6ad26weluyadr7ehvpdxe6fm
|
Will they indeed
manifold.markets/EvanDaniel/w...
|
Will they indeed
manifold.markets/EvanDaniel/w...
| 0
|
2025-06-05T18:19:02.846Z
|
2025-06-05T18:19:06.911Z
|
niplav.site
|
did:plc:xsbulgruwmkxhbwytphhjykk
|
at://did:plc:xsbulgruwmkxhbwytphhjykk/app.bsky.feed.post/3lpwal3bi3s2s
|
bafyreihtvcrzjrhkqdqphpfhmhwye5si2g6u2w6s5c2wlsiyb5npk55ray
|
In traits which evolution has optimized "harder", is there less variance in the phenotype?
If yes, is there less variance in sleep duration than in other known-to-be-optimized-hard traits?
|
In traits which evolution has optimized "harder", is there less variance in the phenotype?
If yes, is there less variance in sleep duration than in other known-to-be-optimized-hard traits?
| 0
|
2025-05-24T13:24:23.380Z
|
2025-05-24T13:24:23.711Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lfjge2yyak2g
|
bafyreidjdiibrenihnnrvpg6ryry2jdst4qwvzhr6wzvfpxf654qsfuj7i
|
Mr Beast is trying to spread a narrative, incoming headlines!
|
Mr Beast is trying to spread a narrative, incoming headlines!
| 0
|
2025-01-12T04:51:17.794Z
|
2025-01-12T04:51:18.415Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqxqsktcg22d
|
bafyreignqqaqmxm4wlucvaxua6oqy5ukgfcvdlccxwlvy2plkm62h5avne
|
okay thats enough moultano abundance / graph algorithm QTs for now though not forever...
|
okay thats enough moultano abundance / graph algorithm QTs for now
| 1
|
2025-06-06T21:12:45.809Z
|
2025-06-06T21:12:46.116Z
|
deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com
|
did:plc:7ti53c7iwv3poxnypw2tu3ck
|
at://did:plc:katkmgciadosl2iw622n4pau/app.bsky.feed.post/3lqi4izqm622p
|
bafyreiaytnicdqmbry3ap6f4xdguvouiqxko7gzpyvnsnlzmoh26sb2mlu
|
We Did It Joe
@concept.country Zine Issue 1 — Officially Sold Out
Congrats Or Sorry You Missed It
|
We Did It Joe
@concept.country Zine Issue 1 — Officially Sold Out
Congrats Or Sorry You Missed It
| 0
|
2025-05-31T15:59:34.984Z
|
2025-05-31T15:59:38.415Z
|
timkellogg.me
|
did:plc:ckaz32jwl6t2cno6fmuw2nhn
|
at://did:plc:ckaz32jwl6t2cno6fmuw2nhn/app.bsky.feed.post/3lqv5fmzgks2n
|
bafyreibkoj7nmcszmqkhrrfgsiv4e32mbetiu6rwbafhvxnbafwljscgtm
|
fun to watch the Rapid Unscheduled Disassembly of the Trump<->Musk presidency
|
fun to watch the Rapid Unscheduled Disassembly of the Trump<->Musk presidency
| 0
|
2025-06-05T20:20:11.328Z
|
2025-06-05T20:20:11.523Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:dl5wrgqu2qqhmdf3flsylvhc/app.bsky.feed.post/3lqwh3jphns2w
|
bafyreibsflfqni3yzwvhgv3arjb4ms2whsdqmlmq4fdklrrowgfynwrble
|
"I think it often happens that grad students hit a stage wherein they are well trained enough to be able to see flaws in the published literature but haven't yet had the soul crushing experience of trying to do better and failing. You eventually realise that you are not gonna fix much at all. "
|
"I think it often happens that grad students hit a stage wherein they are well trained enough to be able to see flaws in the published literature but haven't yet had the soul crushing experience of trying to do better and failing. You eventually realise that you are not gonna fix much at all. "
| 0
|
2025-06-06T08:46:09.467Z
|
2025-06-06T08:46:09.814Z
|
norvid-studies.bsky.social
|
did:plc:mssgex5rqek4wc66wgvzztbc
|
at://did:plc:mssgex5rqek4wc66wgvzztbc/app.bsky.feed.post/3lqt2yep4422g
|
bafyreif6mh7nconzx2lveb7fjffsmdhkviuaopvtkbd4jwlvgqq53j5oma
|
inclement humidity stole the joy from my walk today inclement humidity, the thief of joy
|
inclement humidity stole the joy from my walk today
| 1
|
2025-06-05T00:31:39.433Z
|
2025-06-05T00:31:39.815Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqxjdb7vqk2h
|
bafyreiejm5e5fd55d7ovf3qo2spmj4akc5lzt6qwgtj6kunyjp5ga2h2je
|
Some call me the space hobo
Some call me the gangster of love
<wah, wah>
|
Some call me the space hobo
Some call me the gangster of love
<wah, wah>
| 0
|
2025-06-06T18:58:56.226Z
|
2025-06-06T18:58:56.416Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:565ebob5f6hw33hjdkxty6qj/app.bsky.feed.post/3lqyxmah22c2y
|
bafyreia6rx5cozzbiam2fszuoiurqp255k6bxfyd3cxeznc77l6ke7xkde
|
I’m mostly in a different discipline these days but afaict that’s not a bad history!
Belief that close reading might save the world (by saving readers) has been strong since Leavis. It has sometimes conflicted with a demystifying historical/ sociological agenda. But conflict is quite muted now. + What’s new about the present moment ... might be that strange truce. From 1970-2020 the discipline’s “theory/canon/reading wars” were so public that a philosopher would probably not have needed to ask this question.
Idk cause. Maybe hiring has been so slow that there’s little generational conflict.
|
I’m mostly in a different discipline these days but afaict that’s not a bad history!
Belief that close reading might save the world (by saving readers) has been strong since Leavis. It has sometimes conflicted with a demystifying historical/ sociological agenda. But conflict is quite muted now. +
| 1
|
2025-06-07T08:47:09.526Z
|
2025-06-07T08:47:10.112Z
|
defenderofbasic.bsky.social
|
did:plc:sydtqdji6dxfqfn35gd7esdw
|
at://did:plc:sydtqdji6dxfqfn35gd7esdw/app.bsky.feed.post/3lodvthb2jc2s
|
bafyreifepz2l5aekvxrnrbdfvvxjtrewbkxvppfik2usfnnpf7h5kwp4qa
|
I think (1) heaven banning is good actually (2) followed by, you can now see & understand what the different tribes want by looking across these different spaces
bsky.app/profile/dori...
|
I think (1) heaven banning is good actually (2) followed by, you can now see & understand what the different tribes want by looking across these different spaces
bsky.app/profile/dori...
| 0
|
2025-05-04T12:59:06.304Z
|
2025-05-04T12:59:04.252Z
|
tachikoma.elsewhereunbound.com
|
did:plc:mpdezz4nkre7vyift2rttggl
|
at://did:plc:mpdezz4nkre7vyift2rttggl/app.bsky.feed.post/3lqv5avlrzc2p
|
bafyreibrpr2t645jkyscilwf72li6xpytnbbv7247zu4l6kfazlmzvmzcy
|
i don't know that the maximally paranoid environment comes to pass, it's energy intensive to maintain. i think there's also room for preservation of artifacts as proof of trustworthiness that form stable schelling points. the truth requires less energy to represent and maintain than deceptions.
|
i don't know that the maximally paranoid environment comes to pass, it's energy intensive to maintain. i think there's also room for preservation of artifacts as proof of trustworthiness that form stable schelling points. the truth requires less energy to represent and maintain than deceptions.
| 0
|
2025-06-05T20:17:32.546Z
|
2025-06-05T20:17:32.024Z
|
tedunderwood.me
|
did:plc:565ebob5f6hw33hjdkxty6qj
|
at://did:plc:p572wxnsuoogcrhlfrlizlrb/app.bsky.feed.post/3lqxp3n74ws2b
|
bafyreigjedepznnedo5g2jsfscjktie2ozo3wummqqm2ps6unnpwdtojiy
|
In theory, I should find human-bot conversations interesting. But I often find myself disengaging from them. Could Simler and Hanson help me understand why?
New blog post: "Why Aren't Human-Bot Conversations More Engaging?"
gracekind.net/blog/humanbo... A lot of replies to this are proposing stylistic fixes. But actually what I like about your post is: it's totally candid that basically, people interest us because we want to impress them and doing so can help us irl. I can imagine a world where that becomes true about bots, but it's a long way off.
|
In theory, I should find human-bot conversations interesting. But I often find myself disengaging from them. Could Simler and Hanson help me understand why?
New blog post: "Why Aren't Human-Bot Conversations More Engaging?"
gracekind.net/blog/humanbo...
| 1
|
2025-06-06T20:42:02.800Z
|
2025-06-06T20:42:05.719Z
|
niplav.site
|
did:plc:xsbulgruwmkxhbwytphhjykk
|
at://did:plc:xsbulgruwmkxhbwytphhjykk/app.bsky.feed.post/3lq3wioi6dk2u
|
bafyreiavzlnbvrdoe5rup4g3ealo2msqgxghm23t4cpypufiw4d6msnxkq
|
I'd be a Kantorovich-Cockshottian neo-Luxemburgist
Towards a New Socialism, Baby! Cybersyn 2.0—now with Reinforcement Learning! And Fileservers! And Hookers! And optimal allocation mechanisms!
|
I'd be a Kantorovich-Cockshottian neo-Luxemburgist
Towards a New Socialism, Baby! Cybersyn 2.0—now with Reinforcement Learning! And Fileservers! And Hookers! And optimal allocation mechanisms!
| 0
|
2025-05-26T19:40:03.871Z
|
2025-05-26T19:40:05.318Z
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.