dave
Everyone's predicting a leap in general purpose text-to-video this year, but the real winner will be whoever does text-to-cat gif
tweet

Ethan Mollick
One of the biggest questions with AI is how it interacts with skills.

For example, adding robots to surgeries doesn't do much for the best surgeons, but benefits bad ones. Low-skill surgeons are much worse than good ones, but robots close the gap by 50%! https://t.co/QmHuINOgwT https://t.co/ixfFHHrjjE
tweet

illustrata
RT @B0rriss: The “Advanced (chatGPT) Prompt Writer” tool is here!

🔸10 Templates (listed down in the thread)
🔸 0ver 250 unique property values
🔸 Super easy to customise to your specific needs

It is a powerhouse!

(and the version 2 is coming soon) https://t.co/h3nsjBAlia
tweet

Stable Diffusion 🎨 AI Art
Sleeping on the clouds #AIArt

#StableDiffusion2 / #StableDiffusion #DreamStudio https://t.co/Wzx0syAz8c
tweet

nearcyan
any means to "watermark" text model output is easily bypassed by simply taking the output and asking a second model to rephrase it

this generally works with other modalities as well
tweet

Ben Tossell
in today's email

https://t.co/ZGFEsdaskU https://t.co/HsTp0h1j0X
tweet

Inworld AI
Today our CEO, Ilya @gelfenbeyn, talks the past, present and future of game NPCs - with an early look at our new report on The Future of NPCs. (Hint: it involves #AI.)

Read the blog and get a sneak peek here: https://t.co/QmzaBqvnco

#gamedevelopment #aicharacters #AIgaming
tweet

proxima centauri b ▫️ open edition live!
GM ✨ #synthography https://t.co/Dgi03Omcnm
tweet

Jake Bryant
What if you could choose different camera apps from the lock screen? @apple https://t.co/etmetZceXv
tweet

Aran Komatsuzaki
DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature

Detects samples from pre-trained LLMs using the local curvature of the model's log probability function.

proj: https://t.co/7S5wK1irV9
abs: https://t.co/sze1MIBHOH https://t.co/pQ0ZsSabSY
tweet

AK
simple diffusion: End-to-end diffusion for high resolution images

achieve state-of-the-art on image generation among diffusion models without sampling modifiers on ImageNet

abs: https://t.co/iWZPkFPJq7 https://t.co/Y8z9S3DJR0
tweet