A million things → what's happening in AI, summarised.
Here's a bunch of cool shit that's going on.
@runwayml released their AI Magic Tools; green screen, inpainting, motion tracking, super-slow motion, text-to-image, image-to-image, text-to-colour grade, extract depth and frame interpolation.
Phew, that's a lot of magic 🪄.
Danny Postma (who built StockAI.com) released 'Article2Image'. Turn your article into a cover image. https://twitter.com/dannypostmaa/status/1580091257589624832
A fully-fledged AI voice studio called Coqui has released its waitlist. I'm #16450 but we Brits love a queue.
Create voices (or use AI 'voice actors'), emotions, voice cloning and organise dialogue. We can all interview fake Steve Jobs soon! https://twitter.com/coqui_ai/status/1579882811669901312
Cool demo showing how GPT-3 can explain academic papers. The demo showcases 'Attention is all you need' - the paper that started it all!
Paste a link → get the transcription.
YouTube transcription tool on top of Whisper.
Hugging Face is not only a company 'named after an emoji' 🤗. It started as a chat app for bored teens, now it offers a library of Natural Language Processing models - quite the pivot.
They've just released 3 BIG things... https://twitter.com/julien_c/status/1580274425022541824?s=46&t=jnP_510npLc2LSvCnduk8Q
Microsoft brings DALL-E 2 to the masses with Designer and Image Creator. Anyone use Office and seen AI images replace clip art yet?!
A podcast with @saranormous (ex-@GreylockVC investor) who has just raised $101m to fund AI companies. It's so awesome to see funds specific to these areas BUT why not just cap it at $100M, it's a much nicer number.
A community for discovering and sharing AI art and prompts. https://twitter.com/divamgupta/status/1580274233812271105?s=46&t=jnP_510npLc2LSvCnduk8Q
20min video on how Imagen & Parti make the artistic process more accessible.
Synesthetic.ai, search and remix 10M+ Stable Diffusion images.
Another one... Visualise.ai - stable diffusion tool and prompt sharing site
Raoul Pal, former hedge fund manager, came in with this bold prediction:
This video demo shows Luma AI - It uses self-learning AI to guess parts of the missing scan when you're photo scanning. Shoutout to Amit who's building Luma
An article came out yesterday about a ghostwriter making $200k/yr writing tweets for VCs. Alexis Ohanian tweeted the below, which I imagine is already happening!
If you didn't watch the Elad Gil interview with Sam Altman (which i summarised in the last email) here's a transcription (made by Whisper of course!) created by Kazuki.
Reid Hoffman (the other ' The Hoff') digs into the latest on autonomous vehicles, safety and challenges.
AI is now an expert at the games; no-press Diplomacy and Hanabi.
New AI algorithms predict sports teams’ moves with 80% accuracy. Move over Brad Pitt, you're old news, there's a new Moneyball in town.
Input a YouTube link → get a video w/ subtitles (+ txt, .vtt, .srt files).
An auto-subtitled video generator using Streamlit and OpenAI Whisper hosted on HuggingFace spaces.
Whisper can translate 98 languages to English. That's 98 more than me!
🚨 SCARE WARNING at 9 seconds!!! Someone on Reddit was using the Uberduck voice to read a story and a DEMON 😈 started speaking 😱 - I actually did get a little freight. Hope I can sleep tonight.
A pretty epic video of animation and magic.
Ai-Da makes history after becoming the first robot to be grilled by UK's House of Lords
And that's a wrap!
I curate and summarise what's cool in AI in my daily (mon-fri) email.
Subscribe here: bensbites.beehiiv.com