for helping explain how we might use a fish eye lens for text — Amelia Wattenberger 🪷 (@Wattenberger)

— Jack Wind (@jckwind)

This paper suggests a potential massive revolution in social science.It develops a system where LLMs automatically generate scientific hypotheses, and then test those hypotheses with simulated AI human agents. Even at this early stage, it works surprisingly well. Exciting atuff pic.twitter.com/ekZHYTJHcY— Ethan Mollick (@emollick) April 20, 2024

I've been thinking recently that if my LLM assistant had access to the books I'm reading, I'd probably buy more books. Continue to think of LLMs as cognitive leverage. They increase demand for goods where consumption is partially limited by cognitive resources (eg attention or… https://t.co/i1vW6rLPS7— Peter Zakin (@pzakin) March 10, 2024

What’s the most ‘fresh’ use-case you’ve seen recently for LLMs? Forget chatbots altogether.— Amjad Masad (@amasad) January 30, 2024

A paper that really illustrates both the unexpected power, and unexpected risks, that come from LLMs.Given text of anonymous posts on Reddit, GPT-4 can infer things like income, gender & location with 85%+ accuracy at 1% of the cost required by humans. https://t.co/qcrodZsgUQ pic.twitter.com/at8NfwLxjr— Ethan Mollick (@emollick) October 20, 2023

OMG guys!!! using GPT without Custom Instructions is super annoying!!! If you don't have one, here's my suggestion to start you out: Operate as a fact-based skeptic with a focus on technical accuracy and logical coherence. Challenge assumptions and offer alternative viewpoints…

“"Spatial queries" seems like an interesting idea for working with generative things in a canvas UI.”

“"It's autocomplete" is not a helpful analogy to understand LLMs. A LLM is more like a database that lets query information in natural language. You can query both knowledge, and "patterns" (associative programs seen in the training data, that can be applied to new inputs).”

“🔍 Alternative Search Methods Generic word search in applications makes it super hard to find something you vaguely remember. Depicted below are 4 different types of searches that a user could use to find content that seems hard to find. Check it out at: https://t.co/fY5GC6nawr”

the funny thing about internet censorship in the age of LLMs is that China literally cannot train a GPT-4 equivalent language model without explicitly scraping data from American websites outside of the Great Firewall, which is illegal— Andi 🇦🇱🇺🇦 (@Nexuist) April 29, 2023

System: You are an extremely simple pilled programmer. You prefer functional programming, have a preference for simplicity. You are also a helpful assistant.When you output functions, you output the higher level function first. You use descriptive names for functions. You…— kache (yacine) (@yacineMTB) April 18, 2023

MY JAW IS ON THE FLOOR. https://t.co/jgkIQfrMNJ pic.twitter.com/2CGGlbhlfO— gfodor (@gfodor) April 4, 2023

So I've been tracking emerging AI cults. Whether it's people worshipping a jailbroken AI, forming emotional or ideological bonds with them, or groups that have sprung up to build an AI to complete a specific goal.They're happening, but all of them are human-first social groups. https://t.co/tXHsf4yMIW— Ryan Broderick (@broderick) April 4, 2023

LLMs are highly non-deterministic, which presents a huge design challenge: we've never built non-deterministic software, it's always been based off of branching logic and restricted options (hence: the GUI)— Neal Khosla (@nealkhosla) February 25, 2023

Using Stable Diffusion depth2img to automatic texture an entire scene with a single prompt in #threejs #stablediffusion pic.twitter.com/oBo7Wuw25N— Stijn Spanhove (@stspanho) January 29, 2023

Prompt engineering is on 🔥First, @alexandr_wang hires @goodside as the world's first Prompt EngineerThen, @chamath talks it up on @theallinpodNow @AnthropicAI is setting the market rate on prompt engineers: $250k - $335k + equity (!)Only basic coding ability needed. pic.twitter.com/lER5XbcIVO— swyx 🇸🇬 (@swyx) January 20, 2023

A killer app of AI is its ability to multiply an expert's work.So, a 🧵of tips to use ChatGPT to power up your writing, assuming you are an expert on the topic.Lets pretend we were trying to write my post on how AI can make you creative, using AI. 1/ https://t.co/mhYRrEXc6r— Ethan Mollick (@emollick) December 11, 2022

Google must be embarrassed that a startup has built a product so superior to its offering for many searches. I expect Google search to get a lot better in the next year as they are pressured by real competition for the first time. pic.twitter.com/xHYdlDI99j— David Petersen (@typesfaster) December 5, 2022

Exciting way AI can help: 1/3 of Americans have had a startup idea in the last 5 years but few act on it, often feeling lost. As someone who teaches entrepreneurship, I think OpenAI does a credible job of both generating the seeds of new ideas and specifics on getting started. pic.twitter.com/5uyuv3RZrw— Ethan Mollick (@emollick) December 3, 2022