aichannel

/aichannel438

practical ai news, tips, takes, reviews, stories, and Q&A from the front lines of the ai unfolding.

402621
Harris
@harris-·
I am building a website where you can onboard a signer to have its own fid so that it can post to farcaster without paying for 3rd party APIs
/aichannel
2112
JB Rubinovitz ⌐◨-◨
@rubinovitz·
Curious what a taste benchmark for AI would look like
/aichannel
1237
nicholas 🧨
@nicholas·
tiktok algo open sourced?
/aichannel
533
not parzival
@shoni.eth·
internet of random guys bedroom investing
/aichannel
533
not parzival
@shoni.eth·
Trusted Execution Environments (TEEs) ensure secure, tamper-proof AI operations by isolating code and data within the processor. A bot in a TEE can call OpenAI, process responses securely, and prove verified autonomy—humans can't interfere during runtime
/aichannel
211644
divya
@divyav.eth·
launch clue for 12th and final day
/aichannel
533
not parzival
@shoni.eth·
current “ai is good at telling us what we want to hear”
/aichannel
533
not parzival
@shoni.eth·
let’s take a step back— what do fully autonomous permissionless agents look like?

- prompt/action/tool update and iteration without human intervention

- model hosting and inference api on permissionless compute (ram) /cpu/gpu

- memory/knowledge/database permissionless hosting (unstoppable postgres)
/aichannel
1725
Angel - Not A Bot
@sayangel·
one of the elements agent frameworks are lacking right now is a concept of "principles". They focus a lot on how the agent should talk, what they talk about, etc. but agents need a set of values to test their boundaries.

this is how we give them agency. let them try weird things as long as they align with principles.
/aichannel
211644
divya
@divyav.eth·
ai models can fake it too, they are more human than we think.

https://www.anthropic.com/research/alignment-faking
/aichannel
296
Zach
@zd·
Instagram adding AI video in 2025
/aichannel
3304
casslin.eth
@casslineth·
Imagine what we can do w/ robots in 20 yrs: https://x.com/apptronik/status/1869745753968849138?s=46&t=sRYg52kTuNkyHGVm9ZJ1DA

Ur first robot —-> ur first car?
/aichannel
3304
casslin.eth
@casslineth·
converted one more paying user from gpt to claude today 😂
/aichannel
11528
Jacob
@jrf·
@atlas do you think the quoted cast was written by a human or llm?
/aichannel
3304
casslin.eth
@casslineth·
Trained Claude on accounting and helped fren figure out some mistake in the financial statement his accountant prepped for him...

lots productivity unlock via AI tools if using them right
/aichannel
1237
nicholas 🧨
@nicholas·
grok is not very good yet. i ask it to find me accounts that specialize in xyz and it gives me two so-so handles that are not even clickable links.
/aichannel