the roi on gpt-4o-mini is borderline insane just blew through 7M tokens for $1

Someone recently asked me what type of content they should make, they want to boost their personal brand I said, whichever gives you the most energy It’s such a grind to constantly produce (anything) and you can’t lie to yourself for long If you don’t have energy for your work

~9 weeks left The $1.1M prize pool is still in the bank account

Working on creating a fan-site for My First Million that hosts *quality* info extracted from the pod It's tempting to automate 100%, but then you'll get slop * No: Make AI output 100% the final product * Yes: Have AI do 95% of the grunt work, I add the finishing touch

NYU did an independent human study of ARC-AGI tasks and found that 790/800 (98.7%) of public tasks were solvable by a human This is why ARC-AGI is so interesting... How many other benchmarks have high human performance but low AI performance? https://arxiv. org/abs/2409.01374

Pretty wild what an awesome public, capable, spokesperson for company can do What multiple of Alex’s salary has he generated in value for Anthropic? 20x? 50x? He is crushing it

My hypothesis on AI Twitter bots So 90% of the bot replies I see joined in May '23 My guess is that a capable solo indie developer spun up 2-3K of them and basically set them on auto pilot They're still running, they maintain it out of hope of value, but they aren't sure where

maybe a new level of abstraction I don't need will save me

If the model is as good as advertised, pretty wild that it just took "a few weeks" to train and release it How does this tie to years and $B of dollars for data centers? > I’ve been noodling on this idea for months, and finally decided to pull the trigger a few weeks ago. I

Using Tarot Card Flux to create cards for business framework topics Podcast transcript > Pull Framework > Ask Claude to make an image prompt > flux tarot card These two are: * Business acquisition as a gateway drug for entrepreneurs * How to start an audience from scratch

Pretty awesome, congratulations François Chollet! Both ARC-AGI and Keras have huge leverage for impact towards AGI

logan out here carrying the burden of googles entire ai division this is how you give your product a voice sure there are a few product warts, but you can tell he cares and truly wants to make it better his leverage is 75-150x

I lurk a lot of AI twitter Here’s who comes to mind for each model * Anthropic - Alex, Zack W, Dario * Meta - Zuck? * Google Vertex/Studio - Logan, Jeff Dean * OpenAI - Sam, Mira, Brad L, Andrej/Ilya/Greg (I know) * Mistral - ?

I would pay $2,000/mo for AI that made context switching throughout my day 20% easier.

I'd love to see the legal language that protects him to still shitpost

My guess is someone from OpenAI has already done this (even if unofficially) If not and you want to try GPT-N on ARC Prize then here's the template to do it (very easy) https://www.kaggle .com/code/gregkamradt/using-frontier-models-on-arc-agi-via-langchain DM if you need help

for as important as the definition of intelligence is, we talk about it a surprisingly small amount On Measure of Intelligence by François Chollet attempts to formally define it it's the efficiency at which you acquire *new* (outside your training data) skills cool to see Ted

Who wants to see Mark Zuckerberg live next week in SF? 2x extra tickets for sale, face value Center section, ground floor

Got a DM within 5min of this post and they were sold