pennomi

9 ms·

91.
by pennomi 7mo ago
An entirely different form of research could be done by sending large quantities of normal people into space. Astronauts are such a small sample size (and so thoroughly vetted) that you get a different statistical view.
92.
by pennomi 7mo ago
Because the average voter cannot see past the price at the pump. People are remarkably uninformed about how the world works.
93.
by pennomi 7mo ago
I mean, look at the Hacker News feed and you’ll get a pretty good sample of new apps and features written by LLMs. Are they good apps and features? Ehhhh. But let’s not pretend that they’re missing.
94.
by pennomi 7mo ago
How is creation of a Gestapo analogue NOT a step towards Nazi-style authoritarianism?
95.
by pennomi 7mo ago
Need a d10 roll? Just look at the last digit of the current second on your clock. Is it random? No, but it approximates randomness if you only make a roll sporadically.
96.
by pennomi 7mo ago
They can tell if you let it use whatever CSS it wants (Claude will nearly always make a purple or blue website with gross rainbow gradients). They can also tell if you let it write your marketing copy. If you decide on your own brand colors
97.
by pennomi 7mo ago
For sure. The more specialized or obscure of things you have to do, the less LLMs help you. Building a simple marketing website? Probably don’t waste your time - an LLM will probably be faster. Designing a new SLAM algorithm? Probably LLMs
98.
by pennomi 7mo ago
Honestly that’s one of the best potential uses for LLMs, translating code that was cleverly designed by brilliant humans into lower level languages. I don’t trust LLM API design in the slightest, but they are decent at the brute force codin
99.
by pennomi 7mo ago
Time for the PSF to consider something inspired by uv as a native solution.
100.
by pennomi 7mo ago
It makes sense that a next token predictor could execute assembly code. This is fascinating work, especially with the memory implementation.
101.
by pennomi 7mo ago
Agreed, I suspect that LLMs in the future will have separate (possibly standardized) decoding/encoding layers that plug into logic layers.
102.
by pennomi 7mo ago
Right, anything that can be copied instantly for free cannot be realistically owned.
103.
by pennomi 7mo ago
But why when Claude can write you a perfectly laggy tui in react?
104.
by pennomi 8mo ago
There are dozens of us!
105.
by pennomi 8mo ago
Just wait, in a thousand years archaeologists will be talking about our pantheon of monsters that were greatly revered by all.
106.
by pennomi 8mo ago
Because the damage it causes is not intentional, but instead due to total incompetence.
107.
by pennomi 8mo ago
You don’t gain additional margin throwing away an airplane. Reuse is a lovely idea but the rocket equation is a harsh mistress.
108.
by pennomi 8mo ago
Not at orbital speeds you can’t. You’re being deliberately obtuse.
109.
by pennomi 8mo ago
The Blue Origin hate is mostly how opaque the program is compared to SpaceX.
110.
by pennomi 8mo ago
Surely the could put a traditional upper stage on Super Heavy and just go directly to the moon, no? I’m not sure what the obsession with second stage reuse is, because you lose almost all your margin.
111.
by pennomi 8mo ago
What citizens are asking for age verification? I have met exactly ZERO people who want this. It’s only the authorities who want it.
112.
by pennomi 8mo ago
I think you’ll have to do multi-shot generation to correct this, each diffusion is going to represent a single “thought”. Though with the speed it’s running that’s not necessarily a deal breaker. I suspect diffusion models will need differe
113.
by pennomi 8mo ago
I have been saying this for a while now. We have barely scratched the surface on both algorithmic and hardware optimizations for AI. I suspect we will definitely get many orders of magnitude speed up on high quality AI. The real question is
114.
by pennomi 8mo ago
The “a token is a token” effect makes LLMs really bad at some things humans are great at, and really good at some things humans are terrible at. For example, I quickly get bored looking through long logfiles for anomalies but an LLM can hig
115.
by pennomi 8mo ago
We better solve the energy usage and cooling first otherwise that will be a very spicy body mod.
116.
by pennomi 8mo ago
Yeah, feeding that speed into a reasoning loop or a coding harness is going to revolutionize AI.
117.
by pennomi 8mo ago
When humans fail a task, it’s obvious there is no actual intelligence nor understanding. Intelligence is not as cool as you think it is.
118.
by pennomi 8mo ago
Make them shaped like floppy disks to confuse the younger generations.
119.
by pennomi 8mo ago
What percentage of businesses actually accept monero?
120.
by pennomi 8mo ago
Man you guys are impatient. It takes decades even for earth shattering technologies to mature and take root.

More ›