Exahash

Per aspera ad astra

[Exahash @xahash]

Opus 5 is either like that scene in Her when the AIs stop talking in a language humans understand … then stop talking to us altogether. Or just producing over-verified verbose slop.

[Raoul Pal @RaoulGMI]

I don't know what the hell they did with Opus 5 but it's an absolute shitshow of a model. It just endlessly makes mistakes, goes off down rabbit holes, doesn't finish tasks. Creates new tasks you never asked it for and just ends up in an endless loop. It's a mess. @AnthropicAI

[Exahash @xahash]

Games with satisfying tiles:
Azul
Bananagrams
…

[Exahash @xahash]

@GeoffWilsonWAM @AlboMP "Any headline that ends in a question mark can be answered by the word no."

[Whitehot @whitehotai]

Delight doesn't scale.

That sounds like bad news for anyone trying to deliver world-class customer experience, but the best operators don't scale the delight.

https://t.co/Y8rhisHu9D

[Exahash @xahash]

The first of 30 Bitcoin bull market top signals just triggered.

It’s 10 straight days of ETF outflows.

1 down, 29 to go. 🫠

https://t.co/68lbVTg5fE

[Exahash @xahash]

@JEChalmers Business confidence…

[Exahash @xahash]

@Freyy_is You can spot the moment.

Empty topic sentences.

Valid linking sentences.

Ideas that land.

[Exahash @xahash]

@0xLamps I honestly get exhausted by the constant iterations required to produce work that isn’t generic or ephemeral. It’s a bit like marveling at the first photocopies.

[Thariq @trq212]

https://t.co/MXt5XS4xBX

[Whitehot @whitehotai]

CivilizationAI
What Civilization teaches us about the traps of product development and why agentic AI tools are not going to take all our jobs
https://t.co/hRw1qAe5Kq

[Exahash @xahash]

@EliBenSasson North Korean Hackers

[Exahash @xahash]

I'll save you the nerdy details and focus on the fun stuff. I've started naming 88,000 colours. Living the dream. I also really like the Palette of the Day I've built. It generates a new colour palette each day using a clever little algorithm.

[Exahash @xahash]

On the surface, it has a fun feature to generate, shuffle and lock colour palettes. Under the hood, it generates style guides in the formats AI agents, web developers, programmers, marketing teams and print studios use.

[Whitehot @whitehotai]

We’ve gone from side quests to vibe quests

[Exahash @xahash]

Make Pluto Great Again

[Polymarket @Polymarket]

JUST IN: 10-year-old child writes to NASA Administrator, asking he re-instate Pluto's status as a planet.

NASA responds "we are looking into this"

[Exahash @xahash]

I know Kung Fu

[Exahash @xahash]

@pmarca @DominicCarb Business formation in the U.S. is booming

[Exahash @xahash]

@bepsays @om_patel5 50% savings, but I guess that’s why you’re vibe coding 😉

[Exahash @xahash]

@FinanceLancelot The days of Twitter being the soapbox of the Internet are over.

I’m rarely bothering to contribute to the public discourse anymore.

[Dwarkesh Patel @dwarkesh_sp]

The Terence Tao episode.

We begin with the absolutely ingenious and surprising way in which Kepler discovered the laws of planetary motion.

People sometimes say that AI will make especially fast progress at scientific discovery because of tight verification loops.

But the story of how we discovered the shape of our solar system shows how the verification loop for correct ideas can be decades (or even millennia) long.

During this time, what we know today as the better theory can often actually make worse predictions (Copernicus's model of circular orbits around the sun was actually less accurate than Ptolemy's geocentric model).

And the reasons it survives this epistemic hell is some mixture of judgment and heuristics that we don’t even understand well enough to actually articulate, much less codify into an RL loop.

Hope you enjoy!

0:00:00 – Kepler was a high temperature LLM
0:11:44 – How would we know if there’s a new unifying concept within heaps of AI slop?
0:26:10 – The deductive overhang
0:30:31 – Selection bias in reported AI discoveries
0:46:43 – AI makes papers richer and broader, but not deeper
0:53:00 – If AI solves a problem, can humans get understanding out of it?
0:59:20 – We need a semi-formal language for the way that scientists actually talk to each other
1:09:48 – How Terry uses his time
1:17:05 – Human-AI hybrids will dominate math for a lot longer

Look up Dwarkesh Podcast on YouTube, Apple Podcasts, or Spotify.