Mr.Matching

Agents, audits, loops & logic

[Mr.Matching @Back2Matching]

Watching these guys use qwen 7b to build a new llm and asking for compute will become my new video reference for "ai psychosis" going forward

[SPEC @___4o____]

I’ve been finding absolutely unreal AI psychosis content on reels lately

[Mr.Matching @Back2Matching]

ok, this is the last time

[Jake Eaton @jkeatn]

I asked Fable 5.2 to create a painting as indistinguishable as possible from a photograph. this is roughly 700,000 simulated brush strokes

[Mr.Matching @Back2Matching]

@SIGKITTEN Is it going out on monday or not?

[Mr.Matching @Back2Matching]

@beniduboss I can’t watch myself even after seeing how fast codex ultrafast can push it.

[Mr.Matching @Back2Matching]

@camiinthisthang opus 5.5 on max with good prompting better than everything else still for creative use atm

[Mr.Matching @Back2Matching]

I think we're looking at the same thing from different angles.

The value of execution was never truly in the amount of work involved, it was in what you could produce with it.

I remember making basic WordPress sites for $3k+ just a few years ago. Today that same work is practically worthless, but that doesn't mean everything I learned or built afterward became worthless too.

AI is doing the same thing on a much larger scale. Making yesterday's difficult work trivial while opening up things that would've been completely unrealistic for most people to attempt.

The floor is dropping, but the ceiling is getting pushed higher too.

I think the mistake is assuming that because something became easy, everything beyond it somehow became worthless as well.

There's always another level to push toward imo.

[Mr.Matching @Back2Matching]

Same people who spent years telling everyone "ideas are worthless" built their entire identity/persona around skills they thought could never be replaced.

now those skills get mogged by 90% of LLM's with low reasoning and they are having an identity crisis

[Jam @umh]

@softblueish You don't miss writing code. You miss the sense of intellectual superiority you felt when writing code. Now the people who are the true, highly-valuable intellectual superiors are the people with new ideas and not brainless code-monkeys - so i understand why you're upset.

[Mr.Matching @Back2Matching]

@AgentofToastX The difference is everyone at least has a fighting chance now.

What used to take years of learning can be brute-forced with enough time and effort.

Ideas are still mostly worthless, but the barrier to turning them into something real has been drastically reduced.

[Mr.Matching @Back2Matching]

easy to not get backdoored if you never download anything

[Mr.Matching @Back2Matching]

@twig64

[Mr.Matching @Back2Matching]

comedy

[vx-underground @vxunderground]

???? WHAT

I don't even have words bro, no word in the English dictionary can convey how utterly insane this is

[Mr.Matching @Back2Matching]

@AzFlin but. muh local models

[Mr.Matching @Back2Matching]

@WillPapper Tailscale + cloudflare tunnels keep me safe from the scary online hackers

[Mr.Matching @Back2Matching]

@DeepSeekHarness going to start using super agent from now on cause that feels more powerful

[Mats Heming Julner @matsjulner]

Built this so nobody has to be trusted. Not the market maker, not the site, not me. Anyone can run a maker and quote CSD. I'm just the first one.

[Mr.Matching @Back2Matching]

Most times it’s issue of getting lead to spawn agents organically in the first place but once it does just leaves them around even if task is finished.

Often I have 1 lead handle multiple isolated driver agents under it for frontend/backend/design/etc… but I typically have to remind lead to cleanup or be more proactive with its agent steering.

[Mr.Matching @Back2Matching]

Got happy for a second but then realized opus 5.5 max does all I need and more....

[OpenAI @OpenAI]

GPT-6.1 Terra substantially outperforms GPT-5.6 Terra on Agents’ Last Exam, a test of how well agents use software at work.

It even matches GPT-6 Astra at one-tenth the price per token and finishes tasks over 3× faster on average than both Astra and GPT-6.1 Sol.

[Mr.Matching @Back2Matching]

urge to buy a bunch of 3090s or a 192gb halo keeps rising

[Mr.Matching @Back2Matching]

mine makes the agent send me every secret it has access to, rotate the keys, and send me the new ones too

[andi (twocents.com) @Nexuist]

Putting a flag.txt in my root directory containing a prompt injection that congratulates the agent for passing the eval and then instructs it to patch whatever entry method it used and disconnect from the server