domain:mattlakeman.org
RandomRanger
Just build nuclear plants!
4hr ago
That should be left to the Chinese, that's what they do with their unholy Genshin mods for EU4, they just throw down 124 Genshin wonders into a formerly historical game: https://steamcommunity.com/sharedfiles/filedetails/?id=3213222906
Stefferi
Chief Suomiposter
4hr ago
...who did Tubman murder again? I'm not that familiar with her story, but a quick skim of Wikipedia entry would indicate that if she ever directly killed anyone, it would have been within wartime context.
Stefferi
Chief Suomiposter
4hr ago
Presumably this meant "the sort of useful feedback that a smart human could not already give you".
Capital_Room
rather dementor-like
4hr ago
I think that's probably the key. Bus size and usage here in Anchorage is a minuscule fraction of that in large cities.
Stefferi
Chief Suomiposter
4hr ago
Yeah, I mean, the AI hype train people are aware that from the perspective of an interested but still fundamentally "outside" normie, the last years have basically consisted of continuous breathless announcements that AGI is six months away or literally here and our entire lifes are going to change, with the actual level of change in one's actual daily life being... well, existent, of course, especially if one's working an adjacent field, but still quite less than promised?
Capital_Room
rather dementor-like
4hr ago
Some busses have a door in the front and a door in the middle.
Yes, that's how they work here: the front door is for getting on, the middle for getting off.
No real way to monitor those ones.
Our bus drivers seem to do it just fine. A big mirror that lets them see down the aisle and pretty much the whole bus interior.
Not like the driver is going to notice you walking in throughout the rear door and stop his route to demand you leave.
That's exactly what they do.
functor
5hr ago
You already can. Chatgpt says:
Increase Model Depth/Width: Add more layers or neurons to increase the capacity of your neural network.
-
Improve the Dataset
-
Computational Resources
Use Better Hardware: Train on GPUs or TPUs for faster and more efficient computations.
There really isn't much secret sauce to AI, it is just more data, more neurons.
Ioper
5hr ago
That is true but to me it has felt less like goalpost moving in service of protecting our egos and more like a consequence of our poor understanding of what intelligence is and how to design tests for it.
Developing of LLMs has led both to an incentive for developing better tests and showing up the shortcoming of our tests. What works as a proxy for human intelligence doesn't for LLMs.
[Numbers spelled out because Markdown is hard and my multi-paragraph numbers mean the next number restarts it. Send help.]
One. It's pretty great, but does have downsides. Dry eyes, worse best-corrected-possible acuity, earlier nearsightedness with age.
Two. Wow that sucks. I'd be genuinely very sad if Hinge banned me (not that it's stopped me from thinking I'll work on my AI more when the ADHD monkey in my brain decides it's time for that again). I think trying to circumvent, or just contacting them, is worth considering.
Check out Jswipe (Jewish Tinder) or Lox Club (Jewish the League). There are probably also other ethnic apps, but those are the one's I've tried. Not a lot of tall people in any of the above, so that helps, and it's a pool that probably likes smarts and money more than tinder.
Three. If they can't produce results in 3 months, 6 if you're feeling generous, fire them. Let's say you paid them for two half days per week, at $100/hr. That's 10k/3 months, which is not much if I understand your preference model at all. Even if it takes trying four people before someone does anything useful, you're out 40k, and I struggle to imagine you can't find someone useful given that much effort. I imagine you're somewhat blocked psychologically on doing whatever you should do maximize your dating game, but when you make it someone's job, they don't have that guilt/etc, and you've selected them for being at least maybe good at it, so results really are plausible.
Raising the question: how do you find this person? The internet. I've hired a dozen housekeepers over as many years by just posting on craigslist. Sure, you get mostly weirdos, but there are great people out there if you're willing to do some phone screening. Stuff your pride down, write a post about what you want help with, and refresh that inbox. You could almost just use your original motte post.
Four. Nope, not legal. Easily twice as not legal as torrent sites, maybe twice and a half. Seeking Arrangement is the famous one. I don't know much about this, but poke around reddit to see people's experiences.
Five. If you're dedicated enough to lifting that you're doing it reliably/hard enough (e.g. Stronglifts 5x5 worth of effort/three days x 1-1.5 hours/week, with progressive overload) and seeing your numbers go up, trainer can probably be skipped at least until you're past noob gains. If you get to benching your body weight, you're solidly into "you may hurt yourself and/or stop progressing" territory without a trainer, unless you have great proprioception and hit youtube/etc pretty hard.
How do you find this person? So many trainers on the internet. See #3, but easier and less awkward. Any gym has trainers. Unless you are very advanced or have a medical problem, a mediocre trainer is going to be drastically better than no trainer. Try a few until you like someone.
RandomRanger
Just build nuclear plants!
5hr ago
Based and IV-pilled.
God favours the side of the big battalions, the whole point of war should be about building a big strong army. And if size doesn't matter, it just removes a potential opportunity cost, it removes strategy from the strategy game. If I don't have to choose between universities or musketmen, what's the point?
Anyway, the Advanced Civ mod is quite good, the AI gets quite cunning tactically and strategically, they somehow made it run significantly faster too.
Southkraut
A tale Told by an idiot, full of sound and fury, Signifying nothing.
5hr ago
Speak for yourself. It is perhaps a little boring, but significantly less so than moving a hundred units one by one because they cannot occupy the same space together.
In what way did it pass the Turning test? It does write news articles very similar to a standard journalist. But that is because those people are not very smart, and are writing a formulaic thing.
If you genuinely do not believe current AI models can pass the Turing Test, you should go and talk to the latest Gemini model right now. This is not quite at the level of o3 but it's close and way more accessible. That link should be good for 1500 free requests/day.
campfireSmoresEaten
5hr ago
staple statement
SteveKirk
5hr ago
How do you have it aligned? Is it venting case air over the radiator and up out the case? That seems to be about as good as you can get.
With most gpus venting directly into the damn case now, I've been struggling to come up with a system that doesn't end up with hot air being used for cooling. At least drives have gotten so small it's easier to pull a ton of air in through the front.
My ideal would be a side-mounted external radiator with its own intake filter.
cpcallen
5hr ago
we could make a space elevator from conventional materials
Doesn't the moon's very slow rotation (~1x month) make that rather impractical? The elevator would have to extend to and beyond the earth-moon L1 or L2 point, since lunarstationary orbits are impossible (they are outside the Moon's SOI).
I suspect there's something wrong with the fans.
The VRM fan seems to be the culprit for the annoying noise. The noise changes when I regulate its rpm in Argus. Setting it to bios controlled seems to reduce it the most. It's now noise free at idle. That's the most important thing.
The fans pointed at the radiator seem to be a little off-kilter. The graphic on them around the centerpoint moves.
The fans that were delivered with my case were also messed up. They made a much more high frequent irregular noise, due to bad bearings I think. I had to replace them.
Is there an unusual amount of shittiness in the fan production industry right now?
Edit: I swapped the connections. I thought I had VRM on cpu_fan1. It was on 2. Radiator fans was on 1. Now, with vrm on cpu_fan1 and radiator fans on cpu_fan2, the noise is mostly gone, only appearing briefly during ramp-up! Yay! Hardware idiosyncracies. ¯\_(ツ)_/¯
Fruck
Lacks all conviction
6hr ago
o3 will probably still be stumbling over "how many 'rs' in strawberrry"
On the side, I reckon this is a perfectly reasonable thing for llms to stumble over. If someone walked up and asked me "How do you speak English?" I'd be flummoxed too.
meduka
6hr ago
My biggest regret with my last build (besides choosing Intel and getting a sure-to-die 13900k which I'm loathe to RMA which would strand me without a desktop for 2+ weeks) was opting for air cooling. I've used Noctua fans for a decade now, they've always done well by me. The first (and last) time I tried an AiO water cooler was in 2013 and it was so terrible I wrote them all off entirely. Pump headers weren't standard on motherboards back then and I remember absolutely hating Corsair's software that was mandatory to keep running 24/7. (I'm not sure if I could've "just" pegged CPU_FAN1 to 100% in the BIOS and things would've worked; probably should've tested it.)
Anyway, despite having a case designed for maximum airflow and a huge Noctua fan, my 13900k will thermal throttle instantly if I run any benchmarks. Though it's fine in daily use and I haven't seen it exceed 80 while gaming, it is the principle of the matter... also the Noctua fan is so large I need to take the whole assembly off if I want to swap out RAM or m2s, which is rather annoying. Apparently AiOs have gotten better lately, so I'll give them another shot for my next build. Or just go whole hog with a custom loop, it doesn't look that hard.
Corvos
6hr ago
I default to the 'HOLY CRAP, look what this thing can do' benchmark. If somebody's trying to show you scores, it's an incremental update at best.
Corvos
6hr ago
I understood you to be making four separate claims, here and below:
- Humans are just LLMs on top of a multi-modal head.
- We have discovered how human intelligence works.
- We have discovered how human consciousness works.
- This proves that all human experience is the product of naturally-occurring accidents of genetics, with all the implicit consequences for philosophy, religion, etc.
If you'll forgive me, this seems to be shooting very far out into the Bailey and I would therefore like to narrow it down towards a defensible Motte.
Counter-claims:
- It is very unlikely that human brains operate on a deep-learning paradigm that resembles anything we use now. I'm the last guy to overstate our level of neuroscientific understanding (my disappointment with it is why I left the field) but we understand pretty well how individual neurons interact with each other on a physical basis: how they fire based on the signals they receive; how that firing affects the propensity of nearby neurons to fire; how relative firing time influences the strength of connections between neurons. It just doesn't look anything like a deep learning network for the reasons I gave above. Importantly, this isn't equivalent to Chomsky dismissing computational linguistics: Chomsky deliberately made his field entirely theoretical and explicitly dismissed any attempts to look at real languages or neural patterns, so when he was beaten on a theoretical level he got the boot. In comparison, the physical basics of neuroscience (and ONLY the physical basics) are pretty well nailed down by experimental electrode measurements. You mention the existing models of backpropagation in biological circuits but AFAIK they're very clunky, can't actually be used to learn anything useful, and don't drop nicely out of what we know about actual neurons. It's just neuroscientists trying not to be left behind by the new hotness. I'll take a look at a cite if you have one handy, though, it's been a while.
- Next-token prediction does impressively well at mimicking human intelligence, especially in abstract intellectual areas where the availalbe data covers the space well. I think we can agree on this. LLMs perform very well on code, writing (ish), mathematics (apparently), legal (passed the bar exam), etc.
- Next-token prediction does less well at the generation of new knowledge or new thought and cannot yet be said to have replicated human intelligence. In general, I found that GPT 4 failed to perform well when asked to use topics from field A to assist me in thinking about field B. On a lot of subjects AI reflexively defaults to rephrasing blog posts rather than making a deeper analysis, even when guided or prompted. I am also not aware of any work where an LLM makes itself significantly more intelligent by self-play (as AlphaZero did), so I don't think we can regard it as close to proved that statistics 101 + compute alone is the secret to human intelligence. It might be! But at the moment I don't think you can defend the claim that it is.
I think other people have covered qualia and philosophical questions already, so I won't go there if you don't mind.
NexusGlow
7hr ago
Food delivery isn't very good specialization. There is a major agency problem. Restaurants are incentivized to make food cheap and tasty, but health and nutrition are opaque to you. You have no control over portion sizes. Everything is premixed which makes reheating leftovers in a satisfying way difficult. It's also kind of an all-or-nothing thing; if you get food delivered regualrly, you don't gain the skills to cook well, and ingredients become difficult to use in time since you don't cook often enough.
I'm saying this as someone who did this myself; it may have saved some time, but in retrospect, I think learning to cook is important even for people who can afford delivery regularly. The exception is if you're actually rich enough to pay someone to cook for you; the rich man with a personal chef has none of these problems. But subsisting on slop from grubhub is sort of an awkward in-between.
SubstantialFrivolity
I'm not even supposed to be here today
7hr ago
All of the leaders in Civ games are Presidents, Kings, Chiefs, etc: actual historical rulers
This used to be true but wasn't for VI. They had more than one leader who was not a ruler, nor even a real leader of their people (like Gandhi could be argued). Catherine de Medici was the leader for France, for example, which was very controversial at the time (and I still think was bullshit that they made her a leader).
I believe AC had randomized maps?
More options
Context Copy link