It would be nice to postulate some of these potential emergent systems outlines with timelines. Then it may help better map the granular alignment needs.
Interview question. How would you handle the growing traffic needs and traffic spikes. I'm curious whether any existing architectural diagram of theirs would highlight a potential failure post-mordem.
It's a lot worse. It's like comparing a bow and arrow to an M-16. Yes they both shoot projectiles, but...
The quantity is endless. You can run out of channels, sweep through them a few times, give up, and get bored, while doom scrollers give you an infinite stream of faux-novelty. It's entirely possible to scroll for eternity and never run out of what at least seems, to your limbic system, like information and social stimulation.
The content is generally much lower quality than TV. Even really bad dumb TV is high art compared to a lot of what's on the doom scrollers. It's hyper-optimized for narrow addictiveness and literally nothing else.
There's a closed feedback loop optimization system in place. The site knows you're scrolling. It knows how fast, where you pause, in some cases even where your eyeballs are going. It knows what makes you interact. It knows what you watched before you stopped watching vs what you watched and kept watching. It also has tons of third party data on you and sensor data that often even includes your location. It can optimize its addiction mechanism in a personalized rapid feedback closed loop way that television could never achieve.
It's like a smart drug that tries to keep you hooked. Imagine if the heroin had a learning algorithm that adapted its chemical structure in real time to keep you from quitting and it was powered by fleet-wide learning cross all heroin molecules in the world.
I have it off. Although not backed up I see the image sorta AI scanned where, for a moment it outlines, segments people for a brief moment. Anyone notice this and what is going on?
That runs locally as far as I know. I have backup turned off and I can still search for simple things in the photos (like "dog"), but nothing more complex than that. Only when you enable the backup and it gets uploaded to the cloud do they run more advanced models on it.
From article:
> The marker's notes - seen by the BBC - raise concerns about the "polished/uniform tone" of the work, noting the "consistent flawless grammar and similar sentence structure with no variation in style". The notes also mention a "formulaic paragraph structure that follows a typical AI pattern of 'general claim, the example/result, and conclusion'." Other observations point to repetitive phrasing and a lack of "nuanced critique" as well as a high usage of "em" dashes, a type of punctuation thought to be favoured by AI models.
Judging by the given remarks I see nothing here which proves AI use.
Would you say there is any way to definitively prove a text is AI generated?
I believe it's impossible to prove that some text was generated by an LLM simply due to the limitations of the medium. At the end of the day, text is just a sequence of bytes, and a keyboard can create a byte-identical sequence to an LLM generated one.
Of course you can run various statistical tools, judge a text by gut feeling, or look for specific clues that LLMs usually generate more than humans (e.g. em dashes). But none of those will ever be a proof. Just indications, which together could add up to a very high likelihood.
IMO, the only way to prove an LLM generated some text is to see the original chat session.
I get the privacy concern in regards to advertising ones age, but I wonder if this is akin to a underage walking into a liquor store, interacting with stranger adults, etc. There are obvious appropriate and inappropriate behaviors to minors.
Let's confine only of consumer OS and is provisioned with an underage bracket (you set it to what you want, And Not a Require verify). All sites/app the OS interacts must honor certain child protection laws (privacy, selling, gathering etc)?
What is the main discussion surrounding placing (by choice) some 'optional' age bracket to enter during the OS setup/provisioning? The OS as provider only declares an age and it seems the sites/apps it interacts with to be the ones in the hook, not the OS.