An LLM is a lossy encyclopedia https://
simonwillison.net/2025/Aug/29/lo
ssy-encyclopedia/
…
GENERATIVE AI
-
LLMs as lossy encyclopedias: understanding language model limitations
By
–
-
Gratitude and Recommendation for GPT-5 Diagnoses
By
–
You are welcome and thank you for your kind words 🙂 Best to get possible diagnoses and therapies suggested in advance by GPT-5 reasoning! All the best!
-
Mass Intelligence and the Cambrian Explosion of AI
By
–
The coming mass intelligence—and intelligence explosion, or: the Cambrian explosion of mass intelligence I just read @emollick latest blog post about mass intelligence, an excellent blog post (I'll link to it in the comments, well worth reading), and it got me thinking. Here are
-
Restoring 1920s Paris footage using AI
By
–
Voici à quoi ressemblait Paris autrefois 🗼
— VISION IA (@vision_ia) 29 août 2025
L’heure du déjeuner à Paris dans les années 1920. restauré grâce à l'IA. pic.twitter.com/xyI9ZTfbZCVoici à quoi ressemblait Paris autrefois L’heure du déjeuner à Paris dans les années 1920. restauré grâce à l'IA.
-
Flux Schnell Remains Only Cheap Open Source Image Model
By
–
Flux Schnell, Dev & Pro were launched 13 months ago now, this was a big day for image model space. Since then we haven't really had any good cheap open source models – Schnell is still the only good cheap open source model around, but it is getting outdated. Maybe Schnell v2… https://t.co/2Sl5yQtFwD
— Peter Gostev (@petergostev) 29 août 2025Flux Schnell, Dev & Pro were launched 13 months ago now, this was a big day for image model space. Since then we haven't really had any good cheap open source models – Schnell is still the only good cheap open source model around, but it is getting outdated. Maybe Schnell v2
-
Chatbot Data Privacy: Default Stateless Architecture Prevents Training
By
–
It's very easy to not train on data that people paste into chatbot – that's the default of how they work, they are stateless systems Training on that data requires deliberate engineering
-
Higgsfield implements new AI influencer technology
By
–
Incroyable démo d’influenceur IA… Depuis l’apparition de Nano-Banana, tout va tellement vite qu’il est parfois difficile de suivre la cadence.
— VISION IA (@vision_ia) 29 août 2025
C’est désormais Higgsfield qui l’implémente. https://t.co/UwsZroKhbZ pic.twitter.com/u3dsuDYzCaIncroyable démo d’influenceur IA… Depuis l’apparition de Nano-Banana, tout va tellement vite qu’il est parfois difficile de suivre la cadence. C’est désormais Higgsfield qui l’implémente.
-
Discover SuperMemory: AI-powered personal memory management tool
By
–
Great video, watched it earlier. Btw if you haven’t checked out what @supermemory is doing, you should.
-
AI Struggles with Arrow Recognition on Maps
By
–
I just gave it the map with an arrow. Obviously not perfect (it actually gets the images pretty right, but has trouble reading the arrow of the location). But still, I am very amused.
-
Corporate Data Breach Risk: Employees Using ChatGPT Unsupervised
By
–
Show me any company that can confidently tell you that none of their employees have ever uploaded a private internal memo to ChatGPT for help writing or understanding it