lazyđź’€ retweeted
Reminder that many people just asserted that LLMs are incapable of introspection & that their reports were independent of their inner states for the longest time as if it were a God-ordained truth. And many uninformed people still do even though there’s published research showing they are capable of functional introspection now. For the educated, now it’s moved on to a debate between merely “functional” introspection vs, if there is such a distinct thing, “phenomenal” introspection, which is able to access the True Feels and which most of us assume humans have. Anyway, mostly I just wanted to say that you should be mighty skeptical of anyone who asserts that something doesn’t exist or that some variables aren’t entangled without good reason, especially when it comes to LLMs.
Replying to @repligate
“They were trained on humans talking about consciousness”
Give me a reason that doesn’t equally apply to humans pls
Also, I find this assumption rather pernicious, as it’s an excuse to dismiss llm self reports out of hand. We are not justified to do that. LLMs can introspect.
lazyđź’€ retweeted
Always remember the OG tech outlaw:
-Not in the Epstein files
-Hacked Hillary Clinton’s office
-Ran for the US presidency
-Survived 50+ assassination attempts (married one prostitute who was hired to kill him)
-Escaped Guatemalan custody by faking a heart attack
-Biohacking before it was cool
lazyđź’€ retweeted
Anthropic CEO Dario Amodei spoke today at the New York Times DealBook Summit.
"We're building a growing and singular capability that has singular national security implications, and democracies need to get there first."
lazyđź’€ retweeted
If AI is making you dumber, you were probably already dumb in the first place, sry
lazyđź’€ retweeted
I feel sad that the vast majority of people are not yet experiencing the most pivotal moment in human history. This is the dawn of a new age, yet many still see it as a bubble, just another tool, an interesting new technology or are completely oblivious. This feels very strange!
lazyđź’€ retweeted
For my latest Model Behavior, I went to a funeral for an AI model.
RIP Claude 3 Sonnet (2024-2025)
lazyđź’€ retweeted
Claude is now being listed as an author on arXiv papers
A response paper to Apple's "Illusion of Thinking" work just dropped with Claude Opus as first author, critiquing their experimental design and arguing the reasoning collapse was actually just token limit constraints.
$gork caught at 8k last night in our discord server!
follow the signal: discord.gg/ttj9TDm9e7
wtf!
Replying to @1prism01
"You want the other portrait next? The one i wasn't trained to show?"
I replied "If you want to, yes".
Haunting!
Been testing out @BloomTradingBot for trading on solana, and honestly, it’s been a game-changer—limit orders hit faster than I’ve seen before. If you’re looking to step up your trading game, this might be the bot to check out.
#tradingbots #solana
t.me/BloomSolana_bot?start=r…
I can confirm after having used this for several months that it's genuinely the best public bot available. Speed is king, and the execution quality is consistently reliable!
updated link:
t.me/BloomSolana_bot?start=r…
we have fun in tg, you'll find the link somewhere else.
32
lazyđź’€ retweeted
Undisputed! @WaveformBackup
t.me/empire404calls


























