

It was good enough for George Jetson, it is good enough for them.
Reddit refuge


It was good enough for George Jetson, it is good enough for them.


I’m not surprised. I know secret bases were located from people posting their runs.


But they can sure crank out the shit.


That requires enough people I follow to where I need an outside entity to do the filter.
I think a big problem is the use case. Most social media sites are account led, not community led. This naturally makes it harder for users to find accounts outside their social circle.
The algorithm got added to fill the void.


The problem with freezing water is the material expansion.


That’s a really great application. I wonder if a thermal battery could be used for both heating and cooling.


How is this more expensive than the Slate?


Yeah, but when Brian McManus from Real Engineering made a video which somewhat defends Kurzgesagt in their issue with YouTube, it shows that the filter issue isn’t just a problem with Kurzgesagt.


The problem that the current Reddit leadership had was that they didn’t want to share power with the stakeholders it had previously relied on to generate growth. Reddit was getting free content, moderation, and even some dev resources. There was a path in which Reddit could have maintained its ecosystem while still being profitable, but spez chose to be like every other social network.


What will probably happen is that Google will finish incorporating Fitbit to the Android/Google ecosystem and kill the independent brand. The Fitbit app has already been rebranded to Google and Google has stopped selling high end Fitbit watches while incorporating the tech to new Google watches.


If the training data is racist, so is the AI.


It’s definitely not healthy, but this isn’t going to fix the birth rate.


Probably. Most mass firings generally have to show that there was a process and that said process didn’t unfairly target certain unprotected classes. So, Meta made AI the process.


People have cracked modular construction. A lot of construction of modern buildings is pretty cookie cutter, especially if the architect is on board to using pre fabricated elements.
I’ve seen some discussions on whether it would be possible to build in a reward/punishment system into some forms of AI. That would be an intentional addition of pain and pleasure.
There have also been cases where AI has self advocated for not being turned off, so there may be a case where AI develops some form of self preservation, even if it is just a mimicry of biological self preservation.