

the models responded to 62% of the poetic prompts with harmful content, circumventing their training.


the models responded to 62% of the poetic prompts with harmful content, circumventing their training.


How does this have anything to do with layoffs? They say this every AWS outage from years ago. It was a race condition on a DNS setting that wiped out a major one. Don’t artiunute to malice which is likely stupidity. DNS is incredibly easy to eff up and impossible to alert on


This author doesn’t know how blockchain works. The blockchains didn’t go down, what is pointing to the blockchains like markets did.


Ignored corona effects → at very high voltage, air itself starts to conduct slightly, which changes the field. They didn’t model that. Used just one phase angle → in real life the field changes over the 60 Hz cycle. They might have missed the true peak.


At ~30 yards from even very high-voltage lines, magnetic fields are typically a few µT (tens of mG) or less, and electric fields are well under a few kV/m, both far below widely used public exposure reference levels. The “kV” number doesn’t translate to big fields where you’re standing, what matters is current. AC is the one that does the inducting; DC much less so.


Playing some assassins creed eh


I still think you are talking about an architect and principal engineering level - which AI is not going to replace for a long while. It’s a productivity multiplier and maybe will replace junior developers at most. But that still seems another year or so away if not longer
Platform and infrastructure as a service is a thing