CuriouslyC

6 ms·

91.
by CuriouslyC 2mo ago
Ironic comment of the post.
92.
by CuriouslyC 2mo ago
You say this, but it's likely a lie. Maybe not for you, but for every N people like you that say it, a high percentage of them are lying. There's research on people's preference for AI generated content over human content whe
93.
by CuriouslyC 2mo ago
DeepSeek v4 flash blew the doors off that fantasy. Even if we switched to tokenomics, models are getting good enough and cheap enough that there's no stopping this train.
94.
by CuriouslyC 2mo ago
I hope you posted this ironically.
95.
by CuriouslyC 2mo ago
I think this is a reasonable position, I prefer when people separate their words from their agent's words clearly, but I don't have a problem with reading stuff other people's agents wrote if the input to the writing was real
96.
by CuriouslyC 2mo ago
Would you prefer a 10k word discussion with an agent about a hypothesis with back and forth over data sources and such, analysis, etc, or a 2000 word AI essay condensate with the rough edges smoothed over manually?
97.
by CuriouslyC 2mo ago
You realize that when people aren't told an image or text is AI generated, on average they prefer it to human created content, and it's only when they know it's AI that they rate it lower than the human content? You shouldn&#
98.
by CuriouslyC 2mo ago
I'm looking forward to when AI labs focus more on conciseness of code and writing.
99.
by CuriouslyC 2mo ago
If your job is important enough, it should be worth jogging to.
100.
by CuriouslyC 2mo ago
By definition watermarking narrows and biases the response distribution. Clever algorithms might reduce the perceptual impact and minimize some cherry picked metrics, but it's still worse.
101.
by CuriouslyC 2mo ago
If the challenge is held constant, giving people tools is just teaching a different skill set than teaching them unaugmented. The augmented skill set is more relevant to how people do work in the real world, but working through things unaug
102.
by CuriouslyC 2mo ago
If you're working for someone else, this is true. For people who are building for themselves, it's 100% leadership because managers don't decide what to build or how (at healthy organizations, anyhow)
103.
by CuriouslyC 2mo ago
The combo was glorious. Probably the best shitty pizza I've ever had.
104.
by CuriouslyC 2mo ago
Qwen small models are heavily coding focused, whereas Opus is everything to everybody (even if code is their bread and butter). The downside is they'll frequently hallucinate world knowledge so they need to be RL'd to double check
105.
by CuriouslyC 2mo ago
Model output is pretty mid at augmenting, it can lead to distribution collapse. It's useful for smaller models because nobody wants manually to curate a specialized corpus and those models can't represent the diversity anyhow, but
106.
by CuriouslyC 2mo ago
How to structure experiments/scaling and hyperparameter tuning regimes are most of the secret sauce (besides massive compute). If you don't create an experimental ladder to verify scaling and optimize your hyperparameters well, yo
107.
by CuriouslyC 2mo ago
Small models can be super smart. Big models mostly give you baked in world knowledge, domain flexibility and long context stability/coherence. I wouldn't be surprised if we see Fable level smarts in a coding model that fits in 24G
108.
by CuriouslyC 2mo ago
These results look pretty good, given the smaller model size and the GLM family's historic robustness. Cheaper than Kimi and more robust than DeepSeek. The question in my mind is if you're going cheap, are you going to stop here o
109.
by CuriouslyC 2mo ago
The watermarking is going to get rolled back or Anthropic is going to get rolled. People hate it and it makes the writing worse.
110.
by CuriouslyC 2mo ago
I love lisp, but I don't think the syntax is ideal for LLMs because they generate left to right, and lisp is deeply nested by design. Functional languages that have pipe operators and other features to keep code reading left to right,
111.
by CuriouslyC 2mo ago
I use it on my macbook. The computer use is pretty sweet. It can be controlled from your phone once paired, and it lets you actually see agents that it orchestrates in the threads. Also if you use MCP Apps, it can integrate those into threa
112.
by CuriouslyC 2mo ago
I'd be the first model I'd reach for if I was providing a free service to AI gooners though. Serves them both right.
113.
by CuriouslyC 2mo ago
Ironically, AI mitigates bad developers quite a bit. The architecture/design is coming from above bad devs (in a functional org), and AI tends to write less sloppy code than bad devs, and test/validate it more rigorously.
114.
by CuriouslyC 2mo ago
I don't if it was exactly better, it was less "liquid." Given the reduced competition, you couldn't find primo jobs unless you moved to a hub (which is why most people are in NYC/SF/LA), but when you found a jo
115.
by CuriouslyC 2mo ago
No need for a service, agents with a resume can retool it to match it a target just fine already (talk to a hiring manager to see how much this is already going on) and there are already physical printing + delivery services the agent can p
116.
by CuriouslyC 2mo ago
Unlike 100 years ago, the bald thing is only a negative if you're fat/weak looking. The bald tough guy has been so established in modern lore that even older men can ride the wave if they look/dress the part (through the 50s
117.
by CuriouslyC 2mo ago
My experience is for men that it's not just white collar/blue collar, more it's appearance gives a halo that helps people progress in domains where there's a human, interpretive aspect to advancement. General contractors
118.
by CuriouslyC 2mo ago
I've pivoted to game dev because I'm burned out on AI dev tools. I started working on a hexpunk skirmish roguelite deckbuilder (working name is Hexborne) where you place units and structures on a hex scenario map to build a base a
119.
by CuriouslyC 2mo ago
Blue Prince immediately came to mind for me as relevant modern art. Honestly Myst style puzzle games seem like the easiest type of game for a generative world model to simulate, so I expect if players actually want them we'll see a lot
120.
by CuriouslyC 2mo ago
Agents are getting pretty good at Nix. I think it has sufficient critical mass that it'll be fine.

More ›