One major reason I’m not an AI doomer is that I have a much lower estimate of the direct power of intelligence. This might seem odd, given some of my other content, but let me explain: All the achievements of the human mind only arrive when we cooperate at scale, and involve a lot of iteration and real-world bottlenecks. Newton’s discoveries ultimately enabled us to fly, but they didn’t give Newton the power to fly. If he were isolated, his physics would be almost entirely useless to him. And it’s not just that he couldn’t build an airplane on his own, it’s that he couldn’t design one on his own. Powered flight was the result of tons of iteration, which is to say evolutionary computation. So was the car. Very few of the inventions that revolutionized the world were “one-shotted”. To be sure, having smart nodes in your evolutionary computation makes it faster. Rather than random search as in biological evolution, intelligence at least modestly concentrates mutation probabilities towards more fit forms. This is a big deal, but it isn’t magic. Fundamentally, the world is just not that compressible. Golden motes of compressibility are the treasured exceptions in thingspace, not the rule. I expect the post-AGI future will look much the same. Change will be much faster, but it will still be iterative, cooperative, and bottlenecked by the messy incompressibility of physical reality. Doom makes sense in a world with magic. If intelligence *alone* allowed you to move mountains, I would share the worries. But we don’t live in that world. Here, great things are achieved gradually, iteratively, and cooperatively. Pure IQ does not allow you to levitate or hypnotize or teleport. It makes you a better node in the distributed evolutionary process which is the layer where most of the thinking happens. If I were to psychologize, I’d say people who define themselves by their intelligence overestimate the (direct) power of intelligence.

Relying on these sorts of visual proofs can lead to incorrect conclusions. Here, as we add steps to a staircase it in some sense tends towards the hypotenuse of a right triangle with side lengths 1, but the length of the staircase is always 2, not tending towards √2.

L'auteur du tir qui a touché Zineb #Redouane alors qu'elle fermait sa fenêtre le 1er déc 2018 à Marseille n'a toujours pas été identifié. Elle est morte le lendemain. Lui, c'était un CRS, mais lequel? Appel à témoignage sur @Mediapart (vidéos, photos, etc) blogs.mediapart.fr/edition/a…

Un an de tremblements de terre à travers le monde. Les failles sismiques apparaissent de façon spectaculaire. via @simongerman600

It's that time of term once again for Maximal Monkey Madness! My intro logic students are tossing their bananas.

Motion camouflage! If an aggressor follows the right path while approaching its target, it will appear stationary relative to the environment. As explored mathematically in this paper by @paglend: eprints.maths.manchester.ac.…

What is the P≟NP problem and why would solving it revolutionize the world (and earn you $1M)? A Twitter thread: Contents: ‣ Turing machines and time complexity ‣ Complexity classes ‣ P and NP ‣ Reducibility ‣ Impacts of P=NP ‣ Impacts of P≠NP ‣ Further reading [1/n]

Such a cool way to explain logic gates.

#Hyperbolic #tiling shown in the Poincaré disk model. Geodesic saddle with [3,12] #symmetry and 8-frequency subdivision. en.wikipedia.org/wiki/Geodes…

Cantor's giant softshell turtle grows to 2 metres long & spends 95% of its life buried & motionless, surfacing only twice a day to breathe. (Photos: Megan English)

[2/5] You’ve contributed a change in momentum solely in the y direction, and you’ve caused the particle to speed up along the y-axis. But by increasing its total speed, haven’t you increased its mass-energy—and so increased its momentum in the x direction? How is it conserved?

An inequality of T. Carleman involving the number e = 2.718... as best possible constant. [Reference: N. Dunford & J. Schwartz, Linear Operators, Part I.] #math #calculus

Cynognathus was a dog-like animal closely related to mammals. They were voracious predators & had a global distribution. (Credit: Mark Hallett)

Leaf frogs have an impressive camouflage!

The reason why the shape of a lot of pianos looks like an exponential is because for waves on a string, doubling length means reducing a music pitch by an octave (2x the frequency).

Amazing animation (by @LucasVB) demonstrating how the presence of mass affects spacetime • The bending of the grid lines illustrates the "warping" of space due to the mass at the center • The clocks at each vertex show how time appears to pass more slowly closer to mass