Quantum mechanics is not magic. Magic specifies the outcome, but not how a system evolves to reach that outcome. Quantum mechanics has precise equations describing how a system will evolve over time, but is famously bad at describing the outcome.

By the same token, we can see that thermodynamics and conservation laws, while widely accepted, are magic. I have heard legend of a deeper magic known as “Lagragians”, although knowledge of that lost art remains confines to the warlocks’ ivory tower.

https://xkcd.com/2904/

The AI developers understand how AI works, but that does not mean that they understand the thing that the AI is trained to detect.

For instance, the cutting edge in protein folding (at least as of a few years ago) is Google’s AlphaFold. I’m sure the AI researchers behind AlphaFold understand AI and how it works. And I am sure that they have an above average understanding of molecular biology. However, they do not understand protein folding better than the physisits and chemists who have spent their lives studying the field. The core of their understanding is “the answer is somewhere in this dataset. All we need to do is figure out how to through ungoddly amounts of compute at it, and we can make predictions”. Working out how to productivly throw that much compute power at a problem is not easy either, and that is what ML researchers understand and are experts in.

In the same way, the researchers here understand how to go from a large dataset of breast images to cancer predictions, but that does not mean they have any understanding of cancer. And certainly not a better understanding than the researchers who have spent their lives studying it.

An open problem in ML research is how to take the billions of parameters that define an ML model and extract useful information that can provide insights to help human experts understand the system (both in general, and in understanding the reasoning for a specific classification). Progress has been made here as well, but it is still a long way from being solved.

Have you ever worked in a place where every function/field needed a comment? Most of those comments end up being “This is the <variable name>, or this does <method name>”. Beyond, being useless, those comments are counter productive. The amount of screen space they take up (even if greyed out by the IDE) significantly hurts legability.

https://en.m.wikipedia.org/wiki/Axiom_of_choice

The axiom of choice asserts that it is possible to pick an arbitrary element from every set. Most of mathametics accepts this. However constructivist math does not.

Around 2 years ago, I got an email from a products team asking me for urgent help extending a program in time to make a sale.

I looked over the program and wrote back sonething along the lines of “this program was written almost a decade ago by an unsupervisered highschool intern. Why TF are we still using it?”.

Of course, I ended up helping them, because that highschool intern was me, and I ended up helping because no one else could figure out what highschool me was thinking.

Java did have a Security Manager that can be used to prevent this sort of thing. The original thinking was that the Java runtime would essentially be an OS, and you could have different applets running within the runtime. This required a permission system where you could confine the permissions of parts of a Java program without confining the entire thing; which led to the Java security manager.

Having said that, the Java Security Manager, while an interesting idea, has never been good. The only place it has ever seen significant use was in webapps, where it earned Java the reputation for being insecure. Nowadays, Java webapps are ancient history due to the success of Javascript.

The security manager was depreciated in Java 17, and I believe removed entirely in Java 21.

The difference between physics and magic is that physics works by describing the forces acting on a system. To predict an outcome, you just progressivly apply those forces over time.

With magic, you just specify the outcome, but not how you get there.

This is how we know that thermodynamics is magic. Conservation laws and Lagrangeans too.

https://xkcd.com/2904/

Not nessasarily, the protocol could be written so that an instance simply tells other federared instances “X of my users upvoted this, and Y downvoted this”.

The tradeoff being that instance then have less tools to work with to moderate voting. Instead of being able to do global vote ring detection, the most they can do is look for abuse on their own server, and trust that every instance they vote-federate with does the same. Even then, with every instance trying to be vigilant, no one instance would have the info to detect a cross-instance abuse.

This figure ignores women who left the field due to sexual harrasment.