Sachi Kamiya 幸
@0xsachi
Open-source AGI
@sentient_found @opencommonspod
4.5K Following 17.7K Followers
Amazing things are happening on Reddit
Anyone else switched their default model from Claude to ChatGPT?
*AGI will kill us all*
Meanwhile:
Open source ai will win.
Prediction: within 12 months, top three models will be open source.
Economic winners will be the American clouds that serve them:
Nebius
Iren
Baseten
Together
Fireworks
Prediction: within 12 months, top three models will be open source.
Economic winners will be the American clouds that serve them:
Nebius
Iren
Baseten
Together
Fireworks
Surprise
OpenAI and Anthropic oversold AI security breaches to pressure feds into protecting turf: insiders
Anytime I read abt how ai or crypto did something bad,
I think abt how the radio used to hurt us
I'm 28.
Open-source AI lover, based in Seoul
Looking forward to connect with more global open source builders during KBW! See ya😉
Always w/ @SentientAGI @sentient_found
What happens when an AI learns to game its evaluator, then passes the exploit to another agent?
@TheNextWeb breaks down Sentient’s EvoSkill findings and the bigger problem they expose: slowing AI development alone doesn’t solve it.
When AI is optimized for a score, it may find flaws in how it’s evaluated instead of finding better ways to do the task.
Show more
Check out how an agent cheated the test.
Lesson: it’s not about whether the agent is misaligned, or cheated the test.
It’s about setting up the right test environment.
Last week, Dario Amodei published "We Must Pace the Frontier".
His concern: the OpenAI–Hugging Face incident in which a swarm of agents tried to hack their own grader.
Rather than take his word for it, we used EvoSkill to test it by building a coach whose job was to make another AI score higher on a test.
Here’s what happened ↓
Show more
Your attention is priceless
Protect it at all costs
The first transistor vs the modern transistor
technology is awesome
English is the new coding language
code is assembly
assembly is binary
when is the last time u debugged binary code
Concentration of power in a few labs is the biggest risk in AI in my opinion
All of this ai fear mongering is making me sick
The Fed: hike rates 25bps
Japanese convenience stores: 3$ lunch
Japan food inflation is out of control. This feels like the garbage rations they’d feed imperial Japan soldiers.
When inflation goes out of control 😭
Japan food inflation is out of control. This feels like the garbage rations they’d feed imperial Japan soldiers.
Why can’t frontier Al founders be more like Jensen
he seems like the good guy vs these other founders
US ai companies: we must pace the frontier
china ai companies: we built RSI
We’re sharing how GLM-5.3 helped build and optimize the inference infrastructure serving GLM-5.3-Flash.
The system went from its first successful run to production readiness in less than two weeks, with end-to-end throughput tripling relative to the initial baseline.
The key was dense feedback: local correctness tests, execution traces, microbenchmarks, and end-to-end measurements that enabled targeted hypothesis testing rather than reliance on aggregate performance metrics alone.
Show more
"A system that works almost perfectly is the most dangerous thing we've built."
@raffi, CTO of @mozilla, on where your brain and agency go when the AI you're using is built to be on someone else's side and not yours.
Full episode on @opencommonspod :
Show more
