This is basically what I said, but there are nuances. If a net of 100 lives are saved but there are a net of 100 more people who are put in a lifelong coma and have to be cared for round-the-clock in intensive care until they die of old age, and everything else is equal, I wouldn't count that as a clear-cut "win" or even a "win" at all. Depending on the net value you put on a life of someone living in a decades-long coma, you might count it as a win and that's fair. But I wouldn't.

That's why you run pre-production-quality hardware/software in test environments, and you plan for failure even in production environments.

Who cares if it can be used to build bombs, nuclear and thermonuclear power are and natural so they must be good for you.

Take AI-assisted self-driving cars. SUPPOSE for the sake of argument that self-driving cars CAUSE fewer accidents than they prevent, CAUSE fewer deaths than they prevent, CAUSE fewer serious injuries than they prevent, and the medical and other dollar costs of the accidents they CAUSE are less than the money saved by the accidents they prevent, it would be an overall win, EVEN IF some people die who would otherwise not die.

We don't have enough data to say if this supposition is actually true, but suppose it is.

The same is true for the "seatbelts save lives" argument: It's a net win, even if there are some edge cases where the use of a seatbelt may kill someone who might have otherwise survived.

Ditto medicines. Even for some common medicines that have been around for decades like penicillin, we know that overall they save lives but every now and then someone has a fatal reaction, killing a patient who was not at risk of dying.

IMHO, there are still too many unknowns with AI to let the industry go "full steam ahead" without putting in automated and human guardrails.

It probably was radioactive, just not harmfully so.

Speaking of radioactive things, I have this shipment of bananas....

Ever since deep-fakes became practical, artists and others should know they may be asked to show their work.

These days that means keeping all intermediate materials and revisions. If possible, time-stamp them in a way that is very resistant to being accused as being faked. There's 20th century precedent for creating immutable, time-stamped records.

Should artists have to prove a negative - prove they didn't use AI? Whether they should or not, the reality is these days they do, and any artist paying attention to the world should know this by now.

If the seller knows the chip cannot be used for all of the normal purposes a typical buyer would expect it to be used for, that's a material defect. Failure to disclose could be grounds for a refund and for additional damages if his other components were blacklisted as a result.

Now, if the buyer got it third hand and can't prove the person he bought it from knew or should have known of the problem, he will be out of luck.

It's SpaceAmericaNet.

for warfare.

Log reviews are the kind of thing that an automated tool would be good at pointing out things that "don't look normal" for further human review.

Unfortunately, a well-skilled adversary will know to only do things that look "normal enough" to get flagged as "normal." Doing so will fool even a skilled human reviewer if they aren't specifically looking for the adversary's specific technique.

MONTHS-LONG? Heads should roll and everyone left with any data-security-responsibility will need to be retrained.

Hours or even days-long I can see. But months? Really?

... there were some storage facility for megalomaniacal computers that are self-aware.

The idea of a distributed ledger that can be done worldwide without any government involvement has value by itself.

The idea that you can buy, hold, and sell bitcoin somewhat anonymously* has value by itself.

* obviously trades involving the same wallet are tied together, and there are off-chain methods of tying a wallet to a real person, but with good "opsec" using bitcoin can be about as anonymous as using paper/coin currency or lumps of shiny metal or uncut jewels

But expensive.