Engineer & scientist at the intersection of protein engineering, immunology, and AI | Hubbell Lab | All views are my own and do not represent my employer.
I would be so curious to see how survivors would discuss this. If everyone survived, I suspect many red voters would claim they voted blue. If not, there’s not much to discuss anyway lol
Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?
@noampomsky also there’s a psychological selection effect:
we only get stuck in the things we don’t know how to get out of
if we fall into something we know how to get out of, we get out and it’s a non-issue
so it’s normal to always feel like we’re stuck in things we can’t yet get out of
It is pretty cool that you can now easily write your own software. I made a pdf book to spoken text pipeline, where I play the audiobook in a custom player on Android, where I can press a button that will store the text of the passage playing as a note in Obsidian. So useful!
@asymmetricinfo In addition to this, I often ask for an excerpt that I can use the locate the relevant parts of the source. This makes it really quick to verify
> Instead of trying to put an LLM in the "hot loop" of your program, it's usually worth asking an agent to write a deterministic program to do the thing you need done.
Personal rule of thumb: don't use an LLM for something that a deterministic program can do.
I get it, LLMs are exciting, but they don't mean that software ceases to exist. They are fantastic at dealing with human language and ambiguity, but are terrible (by design and for good reason) at repeatability.
To borrow terminology from the book Thinking Fast and Slow, LLMs are "system 2"...slower, more "expensive" (for LLMs, both in time and dollars), but flexible and creative. Traditional programs are "system 1" ..fast and cheap, but inflexible and dumb.
Instead of trying to put an LLM in the "hot loop" of your program, it's usually worth asking an agent to write a deterministic program to do the thing you need done. Since code is "cheap", this deterministic tool can do exactly what you want it to, and doesn't consume tokens on every execution.
(This applies to agents too..I find myself regularly yelling at Claude to stop repeatedly generating the same 30 lines of python to inspect a file, and instead telling it to generate a 3-line shell script wrapper around jq that it can check in and call repeatedly)
exploring xkcd color survey data (~3.2M named RGB colors)—
if you walk down the list of the most-repeated labels, and for each one, absorb all its nearby colors within a particular OKLab ΔE, you eventually end up with a succinct and quite pretty palette.
New Yorkers and tourists are flocking to Bryant Park to get a glimpse of New York City’s hottest bird this spring, the American Woodcock. These elusive, well-camouflaged birds pause during their migration to forage for earthworms, presenting a unique urban birding opportunity.
My bad ... That's what I truly believe... I can't code but I know how to create things that last forever and power the creation to the masses ... I do that with rap all the time, If I could deliver a new software to my supporters or the masses is insane! " I used to pray for times like this
@meodai It may be helpful to have a tip pop up after the interface is still for a certain delay to tell the user to click or hover. This could have a settings toggle and maybe a cue to disable them after three popups to avoid buffing the user.