Newest at the top
| 2026-01-28 00:33:09 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 260 seconds) |
| 2026-01-28 00:29:14 +0100 | <Rembane> | I want a hot spare |
| 2026-01-28 00:28:13 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-28 00:27:11 +0100 | [exa] | -> back to haskell |
| 2026-01-28 00:26:50 +0100 | karenw | (~karenw@user/karenw) karenw |
| 2026-01-28 00:26:47 +0100 | <[exa]> | like, getting identical copies of fresh brains ain't easy for sure |
| 2026-01-28 00:25:24 +0100 | <jreicher> | "Uncopyable" is my paraphrase. The full point is that with software you can produce multiple instances from the one training history. But when you train a biological brain, you only have that instance. |
| 2026-01-28 00:23:58 +0100 | tomsmeding | likes squishy matrices |
| 2026-01-28 00:23:30 +0100 | <[exa]> | jreicher: wouldn't say "uncopyable", more like "gotten too squishy to do much else than thinking" |
| 2026-01-28 00:22:39 +0100 | tromp | (~textual@2001:1c00:3487:1b00:4c4c:3bb8:a5c6:557e) (Quit: My iMac has gone to sleep. ZZZzzz…) |
| 2026-01-28 00:22:10 +0100 | <jreicher> | I really like Hinton's point (other people have probably made it to) that biological brains provide a tradeoff of being uncopyable but very low power consumption. It's a surprising (for me) way of looking at it, but makes sense. |
| 2026-01-28 00:21:59 +0100 | <[exa]> | c'mon guys no one ever had issues from listening to a good advisor with plenty of great ideas coming from whoknowswhere |
| 2026-01-28 00:20:59 +0100 | <haskellbridge> | <magic_rb> For me also the moral, copyright, and corporate control aspects |
| 2026-01-28 00:17:10 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 255 seconds) |
| 2026-01-28 00:17:00 +0100 | <tomsmeding> | it's one of the main reasons I hesitate to use them even for tasks they are appropriate for |
| 2026-01-28 00:16:33 +0100 | <jreicher> | The whole energy consumption aspect is pretty depressing. I really struggle with that. |
| 2026-01-28 00:15:04 +0100 | sord937 | (~sord937@gateway/tor-sasl/sord937) (Quit: sord937) |
| 2026-01-28 00:12:51 +0100 | <tomsmeding> | and the user guide is even one of them! |
| 2026-01-28 00:12:44 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-28 00:12:32 +0100 | <tomsmeding> | that's better than I expected |
| 2026-01-28 00:12:11 +0100 | <__monty__> | While I'm sure an LLM could also tell you, DDG gives me two hits in the first three results with the query `GHC foo @Int`. Burning a whole lot less energy in the process and not being built on nearly as exploitative a technology. |
| 2026-01-28 00:08:41 +0100 | Lycurgus | (~juan@user/Lycurgus) Lycurgus |
| 2026-01-28 00:07:55 +0100 | <tomsmeding> | possible yes, and in any case if we did scare them off I think we were considerate enough; opinions were expressed, but then, that's what you go on the internet for |
| 2026-01-28 00:07:46 +0100 | tromp | (~textual@2001:1c00:3487:1b00:4c4c:3bb8:a5c6:557e) |
| 2026-01-28 00:07:17 +0100 | <ski> | they could just be humorously expressing that they want to go to focus on the links provided, and not be distracted |
| 2026-01-28 00:06:30 +0100 | <Rembane> | Maybe they'll come back. We'll see. |
| 2026-01-28 00:06:04 +0100 | ski | doubts it was scaring off |
| 2026-01-28 00:05:47 +0100 | <tomsmeding> | oh well |
| 2026-01-28 00:05:45 +0100 | <tomsmeding> | well, some amount of scaring was appropriate, perhaps; not sure if scaring them away was necessary |
| 2026-01-28 00:05:03 +0100 | <jreicher> | Oh dear, we really did scare them off. |
| 2026-01-28 00:04:35 +0100 | <tomsmeding> | I think #haskell is not the right place to go if you want to talk about LLMs lol |
| 2026-01-28 00:04:20 +0100 | Guest41 | (~Guest41@137.83.217.79) (Quit: Client closed) |
| 2026-01-28 00:04:14 +0100 | Guest41 | runs away as fast as he can towards textbooks |
| 2026-01-28 00:04:04 +0100 | Guest41 | nods |
| 2026-01-28 00:03:27 +0100 | Googulator | (~Googulato@77-234-89-65.pool.digikabel.hu) (Ping timeout: 272 seconds) |
| 2026-01-28 00:03:01 +0100 | <Rembane> | fix . Shake $ fist |
| 2026-01-28 00:01:30 +0100 | ski | idly ponders corecursion |
| 2026-01-28 00:01:11 +0100 | <tomsmeding> | there's probably a way to make a circular Shake build recipe called "fist" |
| 2026-01-28 00:01:10 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 245 seconds) |
| 2026-01-28 00:00:37 +0100 | <Rembane> | tomsmeding: Now we just need to figure out how to shake fist recursively and we will have the ultimate old-man-yelling-at-clouds sourcery! |
| 2026-01-28 00:00:10 +0100 | tromp | (~textual@2001:1c00:3487:1b00:4c4c:3bb8:a5c6:557e) (Quit: My iMac has gone to sleep. ZZZzzz…) |
| 2026-01-27 23:59:22 +0100 | tomsmeding | shakes fist at Rembane's shaking fist |
| 2026-01-27 23:59:10 +0100 | Googulator39 | (~Googulato@2a01-036d-0106-030a-3891-da7f-f3f3-f997.pool6.digikabel.hu) |
| 2026-01-27 23:58:48 +0100 | tomsmeding | too |
| 2026-01-27 23:58:40 +0100 | Rembane | shakes fist at Google |
| 2026-01-27 23:58:39 +0100 | <jreicher> | Yep, I've been using them for searching. But as soon as I feel like I need to learn an underlying concept that might be new for me, I follow the link to the source. |
| 2026-01-27 23:57:57 +0100 | <tomsmeding> | whereas putting "@Int" in a search engine is guaranteed to fail |
| 2026-01-27 23:57:42 +0100 | <tomsmeding> | LLMs are an amazing search engine: if you want to know what weird GHC extension gives you `foo @Int` syntax, for example, it's essentially guaranteed to be able to tell you "that's TypeApplications", and then you can look up the docs for that |
| 2026-01-27 23:57:14 +0100 | shr\ke | (~shrike@user/shrke:31298) shr\ke |
| 2026-01-27 23:57:14 +0100 | shr\ke | (~shrike@user/paxhumana) (Changing host) |