尬聊

Second place

[尬聊 @HoboJerk]

@omniverse237 @EsotericCD People who don't understand the difference between protected speech and criminal threats should be shot.

[尬聊 @HoboJerk]

@woke8yearold How so? Maybe I don't understand the distillation game

[尬聊 @HoboJerk]

@travis4nh I sold a bit right before the inauguration of Trump second term so I say this as someone who is practical about this

[尬聊 @HoboJerk]

@travis4nh Predictions like this tend to age very poorly. Short term obviously a bear market.

[尬聊 @HoboJerk]

@elaifresh Is there a citation for that?

[尬聊 @HoboJerk]

@TheSpeculator0 Definitely happened to me at a "nice" one. It had some fake AC that was useless.

[尬聊 @HoboJerk]

@plainstriumph @lymanstoneky Getting real Boss Baby vibes

[尬聊 @HoboJerk]

The new strawberry test and it's even more insufferable

[Valerio Capraro @ValerioCapraro]

Claude Fable 5 doesn’t truly understand. And here is a beautiful proof:

The Beninatto-Trombetti test is a translation test for professional translators. It measures the ability to infer context, revise the surface form, and generalize beyond literal mapping.

For example, the correct translation of:

“Solo 3 parole: non sei solo”

is not:

“Just 3 words: you are not alone”

but:

“Just 4 words: you are not alone.”

An LLM that understands the sentence must also update the meta-linguistic claim inside the sentence.

Claude Fable 5 is arguably the most advanced LLM currently available. And yet it still fails this simple test.

LLMs are extraordinary machines for recombining existing knowledge. But they don’t truly understand.

We are still far from AGI.

[尬聊 @HoboJerk]

@JohnLeFevre No idea who you are but you could never waterboard this out of me what are you doing man

[尬聊 @HoboJerk]

@shakoistsLog Lmao why is he posting this

[尬聊 @HoboJerk]

I still don't get how one can be purely defensive. Spotting a bug is step one to both offense and defense.

[OpenAI @OpenAI]

Patch the Planet is our effort to help open source maintainers move from security findings to merged fixes.

We’re working with Trail of Bits, HackerOne, Calif, researchers, and maintainers to bring Codex Security and advanced models into the remediation process, with human review at the center.

[尬聊 @HoboJerk]

@woke8yearold Defections during a shutdown of fable. So over?

[尬聊 @HoboJerk]

@scootykins @atreides_sf Deported

[尬聊 @HoboJerk]

@tszzl You'd probably want a track record of being mildly right about the supposed dangers before killing millions of people by stopping progress

[尬聊 @HoboJerk]

@leepavelich Jerry, this is a smoke test

Dewwww dew dew dewww

[尬聊 @HoboJerk]

@woke8yearold The bull case

[尬聊 @HoboJerk]

@aelfred_D @d08890 Uplifted Monke Version

[尬聊 @HoboJerk]

@007_lunar @nocontextmemes Adding this to all my memes to make you mad

[尬聊 @HoboJerk]

@woke8yearold @viktorg475 @xctlot I've had in person conversations like this which is very frustrating.

[尬聊 @HoboJerk]

@MorrivarGG @stikves @GuardTm Former I agree disagree with the latter

[尬聊 @HoboJerk]

@woke8yearold @Yuchenj_UW It's not tenable long term to ban use if the companies can't recoup some costs for each model... Unless nationalization?

Really curious what's the end game is for 2026