Building infrastructure for self-improving agents @tangle_network
Joined December 2018
- Tweets3.5K
- Following1.3K
- Followers1.1K
- Likes5.1K
Drew retweeted
I'm firmly in the camp that believes that AI will solve the world's problems long before things can get really bad along any timeline.
Many of the world's most serious problems are due to an uneven distribution of education and, consequently, intelligence, across large parts of the world. But intelligence will soon be accessible everywhere and somewhat uniformly distributed. It only takes one generation of curious kids to be educated and motivated to today's best levels by AI, and a whole country changes shape.
To put my position simply, I think people around the world will become so dependent on external intelligence that they will become defanged, deradicalized, and de-weirded. They'll be biased, influenced, and ultimately changed for the better by the models themselves. We'll see AI basically running people's lives around the world in 5-10 years. And it will make people behave, simply by not helping them if they don't.
I also think the logical extension of LLMs being able to write model weights directly is that in a decade they'll be able to rewrite neurons in your brain. If they can make Qwen 6x better at math, then probably they can make you 6x better at math too.
Maybe the procedure will be via neuralink, or crispr, or maybe in time it won't even break skin. But LLMs are going to be able to rewrite people. Bob will go into the machine to get a math upgrade and he'll come out not being an asshole anymore, and people will be like, huh, what happened?
And so I'm an eternal optimist about the future, no matter how crazy it gets in the short term, say 3 to 5 years. Things are going to change a lot. Imagine what happens when people can rewrite their own genes in 10 years. Society will struggle with it.
But in the 15-20 year horizon, I think we're going to stop fighting each other because of AI. And the world will be an astonishingly better place.
Drew retweeted
This Sunday, Nov. 8, organizers are planning a protest bike ride past Jewish-owned restaurants in my district. It falls on the eve of the anniversary of Kristallnacht - when mobs marked Jewish-owned businesses for boycott.
People have every right to criticize the Israeli government and to protest U.S. foreign policy. But those decisions are made in Washington, not in the kitchens of Center City restaurants. Take those demands to the officials with the power to act.
Holding your Jewish neighbors responsible for a foreign government's actions is antisemitism. No one should have to pass a political test to run a restaurant, earn a living, or feel welcome in the neighborhood we share.
Whatever our disagreements, we should be able to say that clearly - and stand with the neighbors being targeted.
Damn Jews control everything but somehow they couldn’t stop this smh
This is a declaration of a “right of return” for “Palestinian kids” to the “land of their forefathers” in modern day Israel. As I’ve documented at @FoxNews Digital, “Free Palestine” means the “right of return” of diaspora Palestinians to modern day Israel and a demographic overthrow of the nation by Muslims over Jews in Israel to rename it “Palestine.” It’s a “one-state solution.”
Ram Dassman
People should always be in control of the future.
OpenAI is against the notion that it would be acceptable for people to be relegated to a role of doing things that “AI systems think makes sense to reward in humans”.
People should not cede control of the future to superintelligence, and that requirement should ultimately constrain AI progress.
Way to go @DarioAmodei
I’ve been using autoresearch techniques to build fast zkVM, it’s still not fast enough..
Seeing the labs work on this is refreshing and important. Soon we’ll treat zkps like we treat RSI.
New Anthropic research: progress on provable inference for large language models.
Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so.
These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers.
For our full write-up, see here: anthropic.com/research/prova…
Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that end, we're open-sourcing FlashProve’s code, here: github.com/anthropics/FlashP…
We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal (rdalal.github.io) and Ben Mayer (@transmissions11).
It seems spot on the nose that Dearborn, Michigan is the type of place it is.. given it is the birthplace of Henry Ford who beyond his contributions to the automobile was also a virulent antisemite.
Some things are inescapable.
americanfreedomlawcenter.org…
Time to build a decentralized compute network where people resell their idle compute for money?
Take #999
BREAKING: app that has been growing 1000x is facing a compute crunch now that it needs 1000x more compute.
Super cool new areas for autoresearch
Introducing Magnetar from Microsoft Research, the first zkVM that is: (1) general-purpose (RV64), (2) post-quantum from standard hashes alone (no curves, no lattices, no Poseidon), and (3) built for recursion.
Despite its conservative security, it is the fastest zkVM: 7.2–8.7 MHz on a laptop, 10–39× faster than existing zkVMs across 15 workloads! An optimized version reaches 9.5–12.8 MHz.
Links in the thread below.
The weaponization of the Palestinian cause is the leading reason they have no state and no agency over their own future.
Until they are not used as some sort of pawn and weapon to distort history and destroy sovereign nations (which is increasing in Europe and America), their condition will not meaningfully improve.
The only alternative their supporters propose is quite literally the genocide of Jews. It is no wonder their cause seeks to complete the goals of Nazis till this day.
Drew retweeted
If the Arabs had won the 1948 war, there would be no Israel, but also no Palestine.
Just a slightly bigger Egypt and a slightly bigger Jordan and they'd have found something else to hate and fight about.