pianopatrick

3 ms·

1.
by pianopatrick 13d ago
I can understand why it was chosen, but I'm not a fan of writing a bunch of yaml.
2.
by pianopatrick 15d ago
Time will tell how the votes will go in November. Right now the thing I'm hearing on podcasts is that a significant group of people who voted for Trump in 2024 might stay home. We'll know in less than two months.
3.
by pianopatrick 16d ago
I would assume that the computer systems in China that control nuclear missiles, as well as our computer systems that control nuclear missiles, are air gapped and therefore not hackable no matter how smart the AI is.
4.
by pianopatrick 16d ago
But, like, how does that smarts lead to beating nations with today's technology? How would an AI stop an incoming nuclear ballistic missile? I've heard on podcasts with MIT professor Ted Postol that based in the laws of physics in
5.
by pianopatrick 16d ago
When you say AI will be powerful, what do you mean? Like in terms of national power, will AI let the US fight and win a war against China despite China's size, manufacturing base and possession of nuclear weapons? Will AI let the US be
6.
by pianopatrick 17d ago
"Black Swan" would be a good name for an attack drone
7.
by pianopatrick 17d ago
I mean, can't cloud kinda solve all of those if you just use the right things?
8.
by pianopatrick 18d ago
That seems like it'll change over the next few years. Lots of people seem to be making video games with AI. Greater supply, same demand, lower prices.
9.
by pianopatrick 19d ago
Well, that's fine. The great thing about them is you can build them anywhere in the world and people can just access them over the internet. It probably does make sense to build them close to the natural gas wells or other sources of p
10.
by pianopatrick 20d ago
I worked on my AI agent that feels like a unix util. This month I changed the way it writes and edits files https://github.com/patrickjh/ssa
11.
by pianopatrick 20d ago
I dunno, why is 90% the limit? Maybe they can set something up in the chips to use 1/10th the power and set something up in the model design to use 1/10th the power and so we use 99% less power than today. After all GPUs were not
12.
by pianopatrick 20d ago
"it's just physically impossible for this to go on for long" I mean, my uneducated guess is that it's probably within the laws of physics for us to have at least 100x as much compute in the world as there is now. There w
13.
by pianopatrick 20d ago
Someone will likely come up with some clever way to make AI use less electricity. There's probably something in how you design the models or chips or some other thing that will be figured out over the next 5 - 10 years to make these us
14.
by pianopatrick 23d ago
If there's one thing that should be obvious from the war in Ukraine it is that relying on private companies is a bad idea for certain types of weapons. Specifically, for low margin weapons that you need in volume. The standout example
15.
by pianopatrick 25d ago
I've had the thought that an underexplored use case for llms is treating them as classification machines. Put in a lot of input then ask for a single token classifying the input. Like "does this code have any bugs? reply "yes
16.
by pianopatrick 26d ago
I've been thinking about a new AI based dimension to this. If your program is split into smaller decoupled "modules" then all the code for each "module" can fit into an AI context window. In this way the AI can hav
17.
by pianopatrick 27d ago
Well, I think one good thing we can conclude from the music world is that there are people out there who want to pay for humans to do things even when the computer could do it better. I.e. when you go to Vegas and go to a bar, you want to h
18.
by pianopatrick 27d ago
I wonder what would happen if you just told the AI to speak in broken english... Hilarity may ensure.
19.
by pianopatrick 28d ago
For all the programmers reading this who think music theory is complex, just realize that is how programming seems to most people.
20.
by pianopatrick 28d ago
If you like that book, you might also enjoy David Huron's books on voice leading and anticipation. I have not read the book you referenced, but from the description it seems like a similar approach.
21.
by pianopatrick 29d ago
Would be interesting / weird / scary if a lot of the benchmark improvement is just AI getting better at cheating. Also I would not be surprised if there are dozens more sites like this that have not been found
22.
by pianopatrick 29d ago
Maybe open AI and Anthropic could just license their models to run on your own hardware. So a fixed cost instead of per token pricing or subscription with limits
23.
by pianopatrick 1mo ago
one random thought I have that is probably wrong is that at some point a black hole has to end. If a black hole is "stuff goes in but does not come out" then if that continued forever, at some point the whole universe would be in
24.
by pianopatrick 1mo ago
I think eventually you need some kind of system that ranks pieces of data based on how useful they are. I.e. for the web we did that with link count etc. We need some other mechanism for judging and ranking pieces of "memory" for
25.
by pianopatrick 1mo ago
Yeah. And if you compare the AI you can run on a high end smart phone today to AI of the 2000s, today's smart phone probably wins. In 10 or 20 years maybe we'll all be running AI models that are currently considered "frontier
26.
by pianopatrick 1mo ago
I think the tech analogy for frontier models is going to be super computers. Super computers keep getting better but most people don't need them for most things.
27.
by pianopatrick 1mo ago
Yeah, well that's the same as all the other AI companies isn't it
28.
by pianopatrick 1mo ago
I use Cursor because I listened to YouTube interviews with the teams behind most of these products and I trusted their team the most.
29.
by pianopatrick 1mo ago
That sounds like more complex pieces of software that likely have security flaws and are easy to misconfigure.
30.
by pianopatrick 1mo ago
You might not need a separate machine for each agent. You could maybe have one separate machine that all the agents run on.

More ›