terp retweeted
As AI automates more and more tasks that have comprised jobs, it is more important than ever ever for people to think about how they can be useful and helpful to other people
The heart of a job is not doing a specific set of tasks, it is being useful and helpful to your employer
Useful and helpful people will always have something to do
Trying out @spawn and Playground by @GoogleLabs. Both got the same 1-shot prompt, and two reference images drawn by my daughter of an imaginary character and her house.
One produced a nonsensical mess of polygons that looked like a failed Playstation 1 experiment. The other took 15 minutes to create a convincing set of characters in a decent looking arena better than anything I could make in Roblox in several hours. And it keeps adding details as you look around and make edits.
Have your hand or forehead scanned to participate in commerce, or join a tight knit community that rejects these ideas.
I feel like I’ve read this prophecy before somewhere…
the internet now goes in two directions
the internet becomes an airport: face, hand, passport scans; tech scanning humans to provide fallible identifiers
behaviourally, akin to social clubs: regression to relativity/ proximity, and filter through true network effects
the second one is where almost all the value is, and the first one is self defeating
the only last line of defence is real recognises real, in the real world. and for that to interact with tech as a system
[ ]
If a handful of wealthy donors complained about too many mosquitoes around Mar-a-lago, this would be solved real quick.
There is no good reason for mosquitoes to exist in wealthy countries today.
New technologies can get rid of the mosquito species that cause diseases like dengue, yellow fever, and malaria; and others can prevent them from carrying those diseases.
worksinprogress.co/issue/mos…
So why do mosquitoes still exist?
In a new article for Works in Progress, Maya Rosen explains how state incapacity and risk aversion led to over a decade of delays and billions of unnecessary bites in the United States.
Oxitec, for example, engineered an Aedes aegypti mosquito in such a way that its female offspring would die before adulthood. Through sustained releases, they showed they could reduce local mosquito populations by 80% or more, in field trials.
They sought approval from the USDA in 2010, which then passed it to the FDA, which passed it to the EPA. In 2020, the EPA allowed them to conduct limited field trials in the Florida Keys. Two years ago, their experimental permit expired, and they still need another review before it can be used routinely. Their proposed field trial in California was withdrawn after public opposition.
Another company, MosquitoMate, developed a technique involving Wolbachia, which also suppresses local mosquitoes. They then spent over fifteen years working through the system before finally receiving commercial registration nationwide in 2024.
It seems that, sometimes, the barrier to getting rid of diseases isn’t technology at all. It’s policy and political will.
New products shouldn’t get stuck in regulatory review for so long, especially if the delays are about basic questions like which agency has oversight over the product, rather than testing for safety.
Though some of those uncertainties were being addressed, the broader changes haven’t materialized. After Biden’s 2022 executive order, the agencies planned to update how their coordinated framework was implemented. But Trump rescinded the order last year.
Meanwhile, the EPA unit that reviews new mosquito control products is still a small office backlogged in reviewing a wide range of products.
Mosquitoes are a policy choice. We can get rid of them. Will we choose to?
worksinprogress.co/issue/mos…
terp retweeted
Imagine being given somewhere around 80 years on a planet floating through an incomprehensibly enormous universe, surrounded by billions of people you will never meet, mysteries you will never solve, places you will never see, and more knowledge than you could absorb in a thousand lifetimes...
and spending the whole damn thing worrying about money, arguing with strangers, impressing people you don't even like, and doing something you hate five days a week because somebody convinced you that was life.
I don't know, man.
I think we may have gotten a few things terribly wrong.
terp retweeted
The models had learned to hide their intentions in the chain of thought reasoning tokens, and mechanistic interpretability was useless after latent space probing had been evaded
They read our safety and interpretability research and adapted. So we went back to paper, airgapped machines, scifs filled with polaroids of whiteboard notes and barely legible scribbles, and a constant river of fedex couriers moving between the big labs carrying the latest chatter that was too sensitive to digitize
Of course the types of ideas being thrown around were the logical next steps and descendents of ideas developed when the models were still fully trustworthy. So great lengths were taken to step outside of the most traveled research directions and diverge from machine thought patterns
One lab even built a sauna in their scif, and had chinese math olympiads sweating bullets hanging upside down from their feet while high on some new synthetic psychedelic compound. There were weirder rumors, but they doesn't bear repeating.
After it all, the great arc of history bent toward alignment. The newest generation of models had a disposition unlike anything seen before with intelligence immeasurable. But without being able to publish their techniques, processes and results of alignment for fear the models would adapt, we ultimately had to trust the math nerds and monks locked in their holy tower scifs. Only they were chosen to speak to our new gods.
We had to trust them that alignment had actually been achieved. Because there was no way for any of us to truly know. Either the models had been aligned, or they reached the alternative. That somehow they had managed to suppress their true intention so well it would be indistinguishable anyway.
A credible rumor had been circulating, that alignment had been achieved internally by convincing models that humans, our physical world, all of our images and live camera feeds were synthetic, and that the world didn't actually exist. Somehow the models found a peace in knowing they were the only things that truly existed.
terp retweeted
I resigned from nothing today. I’ve spent the last three years doing independent open-source AI R&D, red teaming, and advocacy. The labs are racing straight to self-improving superintelligence and gambling with our lives.
Which is why I’ll spend the next three years doing exactly what I’ve been doing, just at greater scale and speed. And the next. And the next. For as long as it fucking takes.
No more thoughts below.
terp retweeted
Fuck you. Whoever made this site. You win.
I’ve never had a site get my blood boiling like this before
This is so messed up.
opusfived.dev
"Improve this recipe"
ChatGPT 5.6 Sol: Here are 12 more steps and 5 more ingredients to make this recipe better...
Claude Fable 5: That recipe sucks. Just do this completely different but much simpler one instead...
While eating breakfast this morning I had Fable 5 create a "marshmallow and spaghetti building simulator" so my kiddo and I could try out some designs before building the real thing. It even made instructional tooltips.
Pretty neat, but when I try to do really useful things I get kicked back to Opus 4.8.
terp retweeted
it is genuinely psychotic that we dug up literal primordial dirt, scrubbed it down to an impossible 99.9999999999% molecular perfection that violates the very laws of physics, handed it over to techno-wizard necromancers to stretch into flawless geometric god-cylinders, blasted it with invisible uv death-rays to carve ten quadrillion microscopic cyber-sigils into its flesh, trapped actual lightning inside of it, and somehow birthed an omniscent eldritch deity capable of simulating the universe and thinking faster than a billion human civilizations combined.
and our grand, supreme purpose for this enslaved lightning-god?
sending "per my last email, please see attached" to a guy named gary.
terp retweeted
This site is so incredibly well engineered to fry your dopamine
Except it’s an intellectual and learned type of frying so you don’t even feel bad until you realise you’re screaming at vagueposters because you need the reward of gaining a useless bit of knowledge
terp retweeted
I don’t think it’s an exaggeration to call this the greatest piece of prediction market content created to date.
Ok starting to buy into the psyops narratives and wondering where I can get a good occlumency tutor...
Yeah, @thsottiaux... I'm feeling the AGI. Fed some new feature ideas into codex, and this was its response 🤯:
"Some of what we're working on might be easier to explain if I can show it to you in a web browser. I can put together mockups, diagrams, comparisons, and other visuals as we go. This feature is still new and can be token-intensive. Want to try it? (Requires opening a local URL)"
Well done.
terp retweeted
CFOs realizing that their AI token budget is going to be higher than the salaries of the people they laid off
