I for one look forwards to our Minimax, Qwen, Kimi, GLM, and Deepseek overlords

After we launched our startup, we had all sorts of folks reach out to sell their GTM services. I went with one group from Vietnam that would make engagement bait Reddit questions with some accounts, and advertise our product in the comments section with others. It was expensive but it worked

Do you think (or care) about the ethics of this sort of behavior? Do you consider it unethical and if you do, under what conditions would you decide to do it anyway?

Reddit is a huge danger to society. There's no doubt that subs about specific non political (and non popular) topics are hugely beneficial, the overall damage the echo chambers do still outweigh these benefits.

That's a huge misreading. Hiding comments in the UI empirically does not suppress discussion, if anything it actually attracts engagement. Lots of people are seeking the "wrong" to "correct" it.

Suppressed debate is almost universally due to biased/captured moderation teams aggressively using bans.

You're wrong, because if your karma fall below a certain number, your comments wont show up anymore. I can show you if you like.

People shouldn't be blocked from commenting because their karma goes negative. Spamming, hateful talk, etc should be a completely different system. Just because what you say is unpopular (in one place mind you) doesn't mean your words should be hidden.

Well, now that it's becoming "the community source" for LLMs it's becoming even more of a target for large-scale manipulation.

Flatpak

I love obscure benchmarks, and I feel like I can trust their results a lot more - afterall, they (probably) weren't benchmaxxed. RuneBench[0] is another good example (how well LLMs can play Runescape)

[0] https://maxbittker.github.io/runebench/

GP literally caught them?

Website doesn't load

This is either generated by ChatGPT 3 or is really cool and I'm sad that I can't understand it

> Simulation Engineering Core Document Identifier: as.oooooooooo.se/paper • Repository Build Vector: Homogeneous Split-Signature Continuum • Verification Status: Verified Stable • [ Sandbox Overdrive ] • [ Qubit Testbed ]

10000% LLM fever dream...

Also this:

> The following production-ready Python processing block is integrated into the telemetry readback array to automatically run the five-point global integrity suite against output log paths:

at this point if you see "production-ready" in anything, it's Claude slop 100%. Also the script is hilarious.

Convenient for OpenAI that China is against big data centers!

I find this analogy particularly humorous, as atomic bombs do not make for good excavators

Hey man, don't ruin my dreams.

https://en.wikipedia.org/wiki/Project_Plowshare

Didn’t expect “Diplomatic issue with Sudan”

I thought I was the only one!

Not because Claude banned my account (but they did that too), but because OpenAI one day decided I needed a phone number to login, and then proceeded to reject my real one.