my favorite kind of pseudo-intellectual activity is reading this, not looking at the graph and telling my friends “you shouldn’t brainstorm with people, look it up”
Our intuitions about creativity are very different than reality. In this survey, most people didn't know:
🧠Group brainstorming generates less ideas than individuals working alone
📦Constraints increase creativity
👩👦Kids are not more creative than adults https://t.co/NyckkiaQXC
hi - team lead behind both projects...@hievalir said it best but for completeness: we try to be super inclusive. we
mentor everyone from the beginning of their rust journey to them being serious rust devs. we teach idiomatic code, good communication and high performance.
there's no contribution that's too small, if anything the opposite, many and small contribs are the best way to get reps in this new exercise. and we can scale up from there.
keen to see you on our "good first issues" on github. feel free to reach out if anything stands out.
why is playing hundreds of different games inherently more valuable than playing one game??? i even agree that this sounds terrible, but this is a very low effort take.
That sounds terrible. That adds up to 4 and half years of 8 hours a day. You could learn the piano in that time, get a phd, learn to dance, write poetry, paint, climb mountains, read hunderds of books, play hundreds of smaller, intriguing games...
@emollick been praying on the downfall of the essay, but i’m more curious of the accuracy of AI essay checkers high school and colleges are already using. how many AI gen’d essays are currently going undetected and what’s the rate of false positives. it might already be dead for some.
@rachelmetz@emollick to me (not a scientist) a major implication is that as computers gain the ability to reason about increasingly complex problems at some point they may be able find solutions to problems we don’t have the answers to.
“We must re-invent homework and develop teaching concepts that utilize these AI models in the same way as math utilized the calculator: teach the general concepts first and then use AI tools to free up time for other learning objectives.”
I just had the most amazing idea.
Imagine you train an LLM with data that is all chronologically labelled. Every book, every scientific paper, every newspaper headline, every primary source from history. Where primary sources are lacking, descriptions of events or archeological finds (carefully stripped of any anachronistic information) can be substituted. Perhaps you pretrain it on some synthetic generic english dataset that has been very carefully vetted to not contain any historical information too.
Training forces it to produce samples from day N having only seen sources from before day N. It must predict the course of wars, fashions, scandals, etc.
If the model is good, it will end up with a tight well-calibrated probability distribution over world events, and you can use the real course of history to validate just how good it is and how accurate it is out to a given time horizon.
You can feed the model fake headlines. Hitler assassinated in 1938. The model will predict what would have happened. Franz Ferdinand survives assassination in 1914. The model will predict what would have happened.
You can run the model up to today and get it to generate the future. You can feed it synthetic headlines representing decisions.
Tell me why this isn't the coolest project ever. 👇
Introducing – Paragraphica! 📡📷
A camera that takes photos using location data. It describes the place you are at and then converts it into an AI-generated "photo".
See more here: https://t.co/Oh2BZuhRcf
or try to take your own photo here: https://t.co/w9UFjckiF2