tesseractcat's recent activity
-
Comment on Cory Doctorow | AI companies will fail. We can salvage something from the wreckage. in ~tech
-
Comment on Cory Doctorow | AI companies will fail. We can salvage something from the wreckage. in ~tech
tesseractcat LinkI respect Cory Doctorow for remaining consistent on his copyright opinions, and I mostly enjoyed this article. If you have the utmost belief that AI will fundamentally plateau somewhere below...I respect Cory Doctorow for remaining consistent on his copyright opinions, and I mostly enjoyed this article. If you have the utmost belief that AI will fundamentally plateau somewhere below humans, then the article seems correct. But I have to pull one quote from the article:
If there had never been an AI bubble, if all this stuff arose merely because computer scientists and product managers noodled around for a few years coming up with cool new apps, most people would have been pleasantly surprised with these interesting new things their computers could do. We would call them “plugins”.
There was never a world in which this didn't become an enormous, international phenomenon. Ever since Turing proposed the Turing test, it's been inevitable. The fact that you can now have a full conversation with a computer is strange and incredible, it would never have been relegated to something exclusively for techies.
Also, this sentence is not entirely true:
Now, AI is a statistical inference engine. All it can do is predict what word will come next based on all the words that have been typed in the past.
This is a common misconception. Modern LLMs are post-trained on reinforcement learning tasks, which means they're not only trained to predict the next word, but also trained to solve problems more effectively. And, in the case of the example used in the article, they are trained to recognize when they don't know something, in order to not hallucinate. Although, just like humans will often misremember things, this isn't perfect (but it has been getting noticeably better with new models).
-
Comment on Tablet suggestions? in ~tech
tesseractcat LinkThe Samsung Galaxy Tab S6 Lite (2022 or 2024 models) can be found for very cheap used/refurbished (often <100$), and has pretty good stylus support (so, good for digital art if you want a really...The Samsung Galaxy Tab S6 Lite (2022 or 2024 models) can be found for very cheap used/refurbished (often <100$), and has pretty good stylus support (so, good for digital art if you want a really cheap standalone drawing tablet). I can't say how it would perform for Foundry VTT or mobile games, but my S6 lite has been pretty good for various art tasks. My one gripe with it is the aspect ratio, I would have preferred a more square aspect ratio, rather than the narrow-ish 16:10.
-
Comment on Libertarianism is dead in ~humanities
tesseractcat Link ParentIf the only natural laws are anarchy, then don't you already have it? No need for a political movement.If the only natural laws are anarchy, then don't you already have it? No need for a political movement.
-
Comment on The fetch()ening (plans for HTMX 4) in ~comp
tesseractcat Link ParentThe HTML standard included application-like capabilities from the beginning using the form element and the iframe element. You might be thinking of XML?The HTML standard included application-like capabilities from the beginning using the form element and the iframe element. You might be thinking of XML?
-
Comment on The fetch()ening (plans for HTMX 4) in ~comp
tesseractcat Link ParentUltimately HTMX is just an extension of existing web ideas. Previously it was link/form -> whole new page, but with a few tweaks it's possible to do link/form/button -> partial new page. Just that...Ultimately HTMX is just an extension of existing web ideas. Previously it was link/form -> whole new page, but with a few tweaks it's possible to do link/form/button -> partial new page. Just that enables 80% of what SPAs are typically used for.
-
Comment on The fetch()ening (plans for HTMX 4) in ~comp
tesseractcat LinkNot much to say, but HTMX is terrific, and really should be integrated into the HTML standard.Not much to say, but HTMX is terrific, and really should be integrated into the HTML standard.
-
Comment on Body time and daylight savings apologetics in ~life
tesseractcat Link ParentWhat system that replaces capitalism would prevent the controversy about whether to have daylight earlier or later in the day? Even in a totally socialist country with worker owned production,...What system that replaces capitalism would prevent the controversy about whether to have daylight earlier or later in the day? Even in a totally socialist country with worker owned production, this would still be a problem.
-
Comment on What have you been watching / reading this week? (Anime/Manga) in ~anime
tesseractcat (edited )LinkI've been watching City (The Animation). It's the next series by the guy who made Nichijou. To be honest, this caught me by surprise. I hadn't really been following the anime adaptation at all. I...I've been watching City (The Animation). It's the next series by the guy who made Nichijou. To be honest, this caught me by surprise. I hadn't really been following the anime adaptation at all. I had read and mostly enjoyed the manga, but the anime is like 10x better. It's possibly one of the best animated anime I've watched, period. Not to mention that they are nailing the timing of the comedy (which can sometimes suffer in comedic manga adaptations).
It sucks that it doesn't look like it's catching on that much. It's obvious that KyoAni is spending a lot of money on the show, so unless it's a big hit it probably won't be getting a second season. It's hard to tell if it's just that the western audience doesn't care as much though. It's definitely a style of comedy that I don't think will appeal to everyone.
I'd recommend City if you enjoyed Nichijou. City is more ambitious with it's scope, it has a much larger cast, with a lot of clever intertwining of storylines. It takes the same kind of approach as Nichijou and uses it to quickly jump between set pieces all over the City, which can be a little jarring, but generally comes together well. The latest episode featured a really impressive simultaneous cut of like 8 different character groups all at once, plus some claymation/stop motion smoothly blended with animation.
-
Comment on Kindle modding wiki in ~books
tesseractcat LinkIf you don't already have an ereader, consider getting a Kobo device, instead of a Kindle. The Kobo ereaders are much more modding friendly, so there's no need to jailbreak or worry about an...If you don't already have an ereader, consider getting a Kobo device, instead of a Kindle. The Kobo ereaders are much more modding friendly, so there's no need to jailbreak or worry about an update bricking your device.
-
Comment on Looking for the PC equivalent of Garage Band in ~creative
tesseractcat LinkUnfortunately there aren't that many good DAWs targeted at novices. If you do want to dive into a DAW however, I recommend Reaper (which is the WinRAR of DAWs, aka it has an indefinite evaluation...Unfortunately there aren't that many good DAWs targeted at novices. If you do want to dive into a DAW however, I recommend Reaper (which is the WinRAR of DAWs, aka it has an indefinite evaluation period). It's not super easy to use, but also not terribly hard to learn, especially if you're not doing anything complex. If you do go this route, I also recommend the synth/plugin Dexed.
-
Comment on Have you made a video game? Can I play it? in ~games
tesseractcat LinkI've made a bunch of prototypes but never got all that close to releasing anything real. I'm pretty happy with what I've put up on my itch though: https://tesseractcat.itch.io/. I'd say I'm most...I've made a bunch of prototypes but never got all that close to releasing anything real. I'm pretty happy with what I've put up on my itch though: https://tesseractcat.itch.io/. I'd say I'm most proud of my concept for a flipnote style heist game, and a somewhat unique 3d modelling/drawing software.
My new years resolution was to put a game on steam but I don't think that's going to end up happening -.-
-
Comment on "The Bullshit Machines" - A free humanities course on LLMs for college freshmen from UW professors in ~humanities
tesseractcat Link ParentI think you're right to be skeptical, and I didn't mean to present Andrej Karpathy himself as an unbiased figure, however I believe that if you watch the video you'll find that it is much less...I think you're right to be skeptical, and I didn't mean to present Andrej Karpathy himself as an unbiased figure, however I believe that if you watch the video you'll find that it is much less biased than the linked article.
-
Comment on "The Bullshit Machines" - A free humanities course on LLMs for college freshmen from UW professors in ~humanities
tesseractcat LinkThis course makes a number of claims about LLMs, mostly substantiated by repeating the claim that "they just predict the next token" which hides a lot of complexity and controversy behind the word...This course makes a number of claims about LLMs, mostly substantiated by repeating the claim that "they just predict the next token" which hides a lot of complexity and controversy behind the word "just". The course also takes a clear stance that LLMs cannot truly reason, understand, or have consciousness when all of these points are hotly debated in the field.
If you're looking for an introduction to LLMs without such a heavy bias I would recommend this great video by Andrej Karpathy: https://www.youtube.com/watch?v=7xTGNNLPyMI.
-
Comment on What is China’s DeepSeek and why is it freaking out the AI world? in ~tech
tesseractcat Link ParentFair, although to defend my point a bit, the hype is still because they replicated OpenAIs o1-style reasoning. DeepSeek v3 had been around for a month or so with little hype. Although maybe that's...Fair, although to defend my point a bit, the hype is still because they replicated OpenAIs o1-style reasoning. DeepSeek v3 had been around for a month or so with little hype. Although maybe that's just because it took a while to accrue publicity, so who knows /shrug.
It would be really interesting to compare with how much money it's costing OpenAI to train their models, unfortunately to the best of my knowledge they're not open with that information.
-
Comment on What is China’s DeepSeek and why is it freaking out the AI world? in ~tech
tesseractcat Link ParentMaybe. From my perspective at least, it seems like OpenAI spends the money to make the expensive breakthroughs, and everyone else copies them 1-2 years later. After all, the hype around deepseek...Maybe. From my perspective at least, it seems like OpenAI spends the money to make the expensive breakthroughs, and everyone else copies them 1-2 years later. After all, the hype around deepseek is entirely because they replicated the new reasoning stuff that OpenAI was working on.
I would be surprised if AGI (not my favorite terminology) wasn't some derivative of LLMs, although probably a more multi-modal model (audio/video/robotics/language/etc), and RL used heavily in training.
-
Comment on What is China’s DeepSeek and why is it freaking out the AI world? in ~tech
tesseractcat Link ParentWhy do you think models will hit the point of "good enough" in the not-so-far future? OpenAI and co explicitly state their goal is AGI/ASI, which seems like an ambitious goal. Also, o1/o3/deepseek...Why do you think models will hit the point of "good enough" in the not-so-far future? OpenAI and co explicitly state their goal is AGI/ASI, which seems like an ambitious goal. Also, o1/o3/deepseek r1 are all starting to use RL techniques, which are very compute intensive.
-
Comment on What is China’s DeepSeek and why is it freaking out the AI world? in ~tech
tesseractcat Link ParentThis is only true if people decide that LLMs at the current level are good enough. Otherwise, more efficient training/inference will just result in bigger models with more capabilities (assuming...This is only true if people decide that LLMs at the current level are good enough. Otherwise, more efficient training/inference will just result in bigger models with more capabilities (assuming scaling laws hold).
-
Comment on I hate 2FA in ~tech
tesseractcat LinkRecently my Google Authenticator randomly stopped generating a few different codes (notably the one for my google account). And I'm not the only one:...Recently my Google Authenticator randomly stopped generating a few different codes (notably the one for my google account). And I'm not the only one: https://support.google.com/accounts/thread/295605525/corrupted-google-authenticator-codes. I always write down the backup codes but 2FA just feels so dangerous to me, like there's no recourse if my phone dies or I lose my written down codes (fortunately this is unlikely). Anyway I don't really have a point but I wish I at least had the option to determine what sort of risk I'd like to take, hacking vs losing codes.
Be careful if you're using Google Authenticator.
-
Comment on OpenAI is a bad business in ~tech
tesseractcat LinkI think this article is interesting, but a bit myopic considering that it's basically missing the biggest part of OpenAI's strategy. They're hemorrhaging money because they're betting that they...I think this article is interesting, but a bit myopic considering that it's basically missing the biggest part of OpenAI's strategy. They're hemorrhaging money because they're betting that they can train a model that is smarter. I think it's uncontroversial that the earnings potential increases the smarter the model gets (very few people would pay for a GPT-2 level model, for instance), so the only question is whether or not they can train a smarter model in time, and what people will pay for it at that point.
One objection is that even if they do manage to train a smarter model, at that point there will be no differentiator from other companies who will also train equivalently smart models. I think this is sort of true, but I wouldn't be surprised if OpenAI can exploit their lead, and then maintain it through network effects.
What specifically are you trying to learn how to do? Generally, I would recommend making an account with one of the big 3 LLM providers (OpenAI/ChatGPT, Google/Gemini, or Anthropic/Claude), and just talking to it. LLMs are pretty good at giving advice on how they should be used.