gubbrora

Consider what happens if neither player wants to go first. P1 passes. P2 passes. Now p1 has to play or the game is drawn. If the game is played with win3 draw1 lose0 rules, then p1 should play even if at a slight disadvantage. A 49 % chance of payoff 3 is better than a 100% chance of payoff 1.

Just elaborating on this. One way of looking at it is that we turn the attacked program into an interpreter thus bypassing the requirement for executability.

It turns out that a lot of programs can be tricked into interpreting a family of languages called rop chains.

Agreed

If we tried to get these thing 100% right we'd not have time for anything else. Punting is pragmatic.

People with clearance have to undergo regular lie detectors. Wouldn't be surprised if a loyalty chip is the next step on that path.

I'm not convinced a very dark future is preventable

Kinda and kinda not. They have done well so far but they could conceivably be beat by a dedicated chip. They need make sure they keep up with the research so they build what the community wants.

Sometimes it really is different. And sometimes it isn't.

There is a subtle false dilemma there. That we have to believe one way or the other.

A better policy would be to assign some degree of belief to both possibilities.

This is how you get stuck in a local optimum.

Communicate that everything worked in a better way.

I think something is lost when doing this. I'd bet the researcher who first builds a model and then reaches for ml will outperform the researcher who goes straight for ml.

Building a custom model will help with feature selection. It will provide a baseline to compare the ml model to which can help debug problem points of the ml model. And finally it serves as a sanity check that you aren't leaving a lot of performance on the table.

No. Per wiki.

Deepfake (a portmanteau of "deep learning" and "fake"[1]) is a technique for human image synthesis based on artificial intelligence. It is used to combine and superimpose existing images and videos onto source images or videos using a machine learning technique known as generative adversarial network.[2] The phrase "deepfake" was coined in 2017.

Because of these capabilities, deepfakes have been used to create fake celebrity pornographic videos or revenge porn.[3] Deepfakes can also be used to create fake news and malicious hoaxes.[4][5]

We've been told that the antidote to deep fakes is supply chain security. It's troubling to hear that supply chain security is already broken and by fairly unsophisticated attackers from the sounds of it.

Is misra as safe as ada? C++ for safety critical systems sounds vaguely concerning

Making 1kg of wolf requires way more than 1kg of prey. A rule of thumb I've heard is a step in food chain has only about 10% efficiency.

If wolves were more prey than predator populations would decline very quickly.