I just published "Final Report for EF Grant FY24–1558 — TrueBlocks MiniDapps"
This is how Ethereum dApps should work.
100% decentralized. 100% private. 100 times faster.
https://t.co/t392Lw2y31
If an AI developer was on track to lose control of its internal agents, would anyone outside the company know?
Today, we are sharing METR’s vision for how we would work with AI developers to openly assess and share evidence about this.
Hear me out the city has been in charge of the former Atwater Kent building and it’s in extreme disrepair. Keeping the collection is fine but you all need to invest in the collection and the building.
@emollick Watching from the outside, the only theory that helps me understand what's going on is:
AIs are not creative. They pick up existing solutions that humans simply hadn't noticed before.
And humans get a peek at the back of the book and push forward through creative action.
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
https://t.co/7N6TPlft1P
@mihai673 Thanks for your concern. Sorry to alarm anyone.
If I send a tx and now must create a new address (due to AI breaking cryptography), keeping track of that will be impossible.
The dream of automated accounting.
@davideoks@amoylan@dwallacewells Fix politics, by which I mean do not allow the politicians to lie, prevaricate, bullshit, separate people, demagogue, or use statistics misleadingly.
@OpenAI I hereby coin a term: "post-conjecture," an era when no mathematical conjectures remain open, only facts and non-facts (Gödel notwithstanding). I call this the Q.E.D. conjecture.
Politicians beware.
@SchopenhauerOn@ObhishekSaha I'd be interested to learn which of the 4,000 problems the AI tried to solve and failed, and why they failed. Is there such a list?
ethereum clients disagree on what trace_* returns, so every explorer, indexer and monitor carries per-client workarounds. i've been fixing that at the source: a draft spec, cross-client tests, almost a hundred upstream fixes merged so far
grateful to @LidoGrants and @LidoFinance for backing the next phase 💙
https://t.co/x2Fsecj6yx
trace_* standardization update
- 86/132 client decisions match the draft spec, up from 44 on sep 25
- 98 upstream fixes merged across 8 repos
- nethermind dev matches every decision we can measure
- all 33 api decisions converged
https://t.co/kBxgkabK9g
My bet is that this extraordinary post took about 1,000 times less time than it would have two years ago, which is why two years ago it would never have existed. This is a master class. We should all raise our bar and be doing masterclass work like this from now on (until it all falls over).
people asked how i made this. the tools were new, but the process was pretty traditional: model sheets, storyboard, key animation, compositing, dailies, retakes.
the target was a restored 1991 anime laserdisc of a film that never existed.