Jay Hack
AI hacker. Tweets about AI, computing, and their impacts on society. Previously did startups, @palantir, @stanford. Not a pseudonym.
View on πThreads
Computer vision had a lot of quick wins in bootstrapping massive datasets over the last decade: π¨ Image colorization - convert color photos to B&W π¬ Image superresolution - downs...
An intriguing trend in AI π€: βModels all the way downβ (aka "stacking") Have models invoke other models, then watch as emergent intelligence develops β¨ Hereβs a discussion of wh...
A simple yet powerful idea emerging for LLMs: Self-guided refinement π¬π§ Get an LLM to critique it's own outputs and iteratively improve. Surprisingly effective. Intuitively maps...
What does an AI sysadmin look like? Introducing π¦ ShellShark π¦ : an agent that swims through your infra to: β Set up/modify infrastructure βοΈ Examine and debug services π» All wit...
Here's my π₯ hot take on ChatGPT Plugins: It's immediately apparent that this is the best channel for an impressive set of AI applications. Several talented folks I know are dropp...
The Matrix is the most important film of all time, ever. Especially now. Itβs a movie about power, agency, and the disconnect between perception and reality. In an era of deepfa...
β‘οΈExciting ideas from tonight's "AI Tinkerers Meetup" 1) LLMs for dementia diagnosis/treatment - Offer seniors conversational partners for company - Use chat to diagnose demen...
This is an EUV Lithography Machine. It costs $150mm, has >100,000 components, took 30 years to develop and is roughly the size of a bus. Now, it's a pawn in a global geopolitical...
How can we cross the chasm from LLMs to intelligent, embodied agents? Thoughts on "PaLM-E: an embodied language model" https://t.co/fn145uSKxs Google teaches a single (562B) mod...
π¨ GitChat - Conversational UI for Pull Requests π¨ https://t.co/5Sp5ydTxMQ Rip through your pull request reviews! πππ Paste in any public PR url from Github β‘οΈ ChatGPT experienc...
π₯Shameo: Deepfakes for @BookCameo Make celebrities say anything you want! Hackathon project from '21; many exciting potential extensions given recent advances in AI Samples & m...
If @ESYudkowsky wants to stop AI, this will help: "Poisoning Web-Scale Training Datasets is Practical" https://t.co/BBZVps4Kh4 Calls out two vulnerabilities in web-scale AI data...