Claude Sends Bad Advice To Laundry

The real news

The story, straight

A saved note is advice left for your future self. If it's wrong, remembering it can make you repeat the mistake.

An AI assistant has the same problem: a bad note can survive into another session.

Anthropic describes 'dreaming': an offline process that uses idle time to check saved notes against original session logs, correct errors and organize linked reference pages. It's not sleep or consciousness.

Think checking yesterday's receipts, not counting electric sheep.

In Anthropic's Pokémon Red demo, Claude saved the wrong location for a PokéCenter near Mt. Moon.

With that note, it went over the same ledge 3 times and never reached Cerulean City. After the process checked and corrected the note, Claude went back into Mt.

Moon and eventually reached the city. Its worst navigation app was its own advice.

In Anthropic's own evaluation, linked notes built from about 350,000 Stainless files improved from 13% to 48% of graded criteria on 25 questions as preparation grew from 2 to 128 sessions. Those scores used the notes alone.

This wasn't independent testing or proof Claude got smarter at everything.

The useful lesson: storing advice isn't enough. Check where it came from before trusting it again.

The process costs computing power; when to do this work remains an open question.

Hardships are coming. I, the Big Dreamer, created this token to help the people.

I truly do see this token going up in value. Ride the rocket with me and let's see where the next 24 months take us.

Buy: https://lilmeme.fun/buy

Repost before your worst advice earns seniority.

Thanks for supporting Lil'meme. Your support is appreciated.