The reality is those parents have tried nothing and are all out of ideas, or, in the vast majority of cases, simply don't care.
This is not at all about children by the way, because all millenials have grown while consuming porn and social media, and haven't turned into degenerates. It is 100% an excuse to spy on citizens, and nothing else.
If it's simple to begin with doing this, or things to try specifically that can build any parents skills and competencies in this area, mind sharing that?
> It generated something like 300-400 megabytes per hour. Despite the insane volume we made this work really well using just grep and other basic CLI tools.
400MB of logs an hour is nothing at all, that's why a naive grep can work. You don't even need to rotate your log files frequently in this situation.
> current models have almost 100% success rate on tasks taking humans less than 4 minutes
The contrary is easily verifiable by everyone individually. It's nowhere near 100%, or even 50% for few minutes tasks even with the best models in real world situations.
I've only noticed that combination (failure of short everyday tasks from SOTA models) on image comprehension, not text.
So some model will misclassify my American black nightshade* weeds as a tomato, but I get consistently OK results for text out from good models unless it's a trick question.
A mutation in our ancestors 10 million years ago likely spread due ground fruit fermenting, becoming toxic to other creatures thus creating an ecological niche. So, even if they were not human it’s reasonable to say the love affair is that old and shared with other species.
“Ten million years ago a common ancestor of humans, chimpanzees and gorillas acquired a mutation that let them remove ethanol from the body more efficiently. This adaptation coincided with a change of habitat. Tropical forests were collapsing, notes Robin Dunbar of Oxford University. Some 90% of apes went extinct. One lineage survived by leaving the trees and foraging on the ground.
Whereas apes in trees gobbled fresh fruit, those on the ground found fallen fruit, which ferments. Thus, our ancestors may have acquired a taste for alcohol–which allowed them to use these scarce calories. This “drunken monkey” hypothesis suggests that a love of the smell and taste of alcohol, the sign of an energy-rich fruit, gave our ancestors an edge. Their chosen poison would have been fairly weak. A study of overripe wild Panamanian palm fruits found none stronger than 5% alcohol—about the same as a Heineken.“
There is that pet theory that alcohlism also converted us from nomads to agrarian societies as mead and bear are impractical to make year round while on the move.
> humanity hasn't existed for anywhere close to 10M years.
From the article:
" Humans, unusually, have a pair of enzymes that turf it out like night-club bouncers. Our ability to process alcohol has deep evolutionary roots.
Ten million years ago a common ancestor of humans, chimpanzees and gorillas acquired a mutation that let them remove ethanol from the body more efficiently. This adaptation coincided with a change of habitat. Tropical forests were collapsing, notes Robin Dunbar of Oxford University. Some 90% of apes went extinct. One lineage survived by leaving the trees and foraging on the ground."
I haven’t read the whole thing, but it starts off talking about a gene mutation in our ancestors species 10 million years ago that lets us process alcohol. So they are taking a little artistic license.
If you calmed down and stopped snapping at everyone, you might understand that I'm writing about how the law and a lack of studies could make some people more willing to drive high. You are substantially diminishing the quality of the discussion here.
The average (presumably arithmetic mean, though it could technically be any of a wide variety of measures) is not particulatly interesting, the median specifically would be more interesting, as a single figure.
I partially agree, but it is still relevant, because there is a relatively low upper bound to the values possible after which someone would literally be unable to even walk to their car to start driving.
When the average is SO high above the legal limit, and with this constraint that there is an upper bound, it's absolutely relevant.
No need to be judgmental about statistics. They are just facts.
A similar result about alcohol would be the (hypothetical) statement that the rate of drunk drivers in fatal accidents was constant before and after the enactment of Prohibition.
I do agree that the fact that fatal THC% stays constant before and after legalization is a surprise.
It absolutely would. If 40% of people test positive for THC, then this would mean there is no effect. I find it unlikely 40% of people test positive for THC, but yes, it does matter.
That wouldn't actually mean no effect, you need 40% of people driving to test positive for it to be no effect. It's unlikely the population driving is equivalent to the population at large - for one there's a set of responsible people who won't drive while high. For another weed use isn't randomly distributed through the population but correlated with certain subsets, which probably have a non-average rate of driving just by coincidence.
(Not that it really matters since I don't buy for a second that anywhere near 40% of people/people-driving are high at any given time. I also don't put much faith in numbers in the abstract of a a yet-to-be-published study...)
There is a case for the two populations to be quite similar.
THC in the blood doesn’t mean actively high for habitual users, which would be most users if THC consumption is high. It means recent use, but not clear impairment.
The article is not saying 40% of all drivers tested positive, it’s stating that 40% of people who died in a car accident tested positive, at pretty high levels too.
The levels described are actually pretty low. The "legal limit" is so low for THC that anyone who's had THC in the previous days could test positive, even if they aren't "high" at the time of driving. It isn't quite the same as the BAC legal limits for alcohol. And it doesn't account for body weight, tolerance, and other factors that definitely contribute to how a driver reacts no matter how long it's been since they consumed THC.
And the study doesn't seem to differentiate between the different types of THC either, some of which are not psychoactive at all and which people use to relieve pain and anxiety. There's quite a lot of people using non-psychoactive THC which wouldn't impair driving.
> It’s stating that 40% of people who died in a car accident tested positive, at pretty high levels too.
It doesn't say anything about the distribution, only that the "average" (presumably, the arithmetic mean, a measure particularly sensitive to distortion by outliers) was at a particularly high level.
Yes, it would be useful. When controlling for variables, you normally want to compare against a baseline.
If 40% of the whole population has THC in them, we'd need a control population (maybe from earlier when thc was less prominent) to see if per capita deaths has meaningfully increased. I'd do the same study, tangentially, for tech workers to see if productivity has changed when controlling for other variables.
That would be true if you looked at a variable which is not influenced by driving, like the percentage that wear red jumpers, but one would hope that not everyone is reckless enough to be highly intoxicated and drive.
This is again THC apologizism, nobody would even begin to suggest this if we were talking about alcohol.
> nobody would even begin to suggest this if we were talking about alcohol.
When we talk about alcohol, we explicitly separate presence from impairment using blood alcohol concentration. We set legal thresholds because studies show a sharp increase in crash risk above those levels, relative to sober drivers. If alcohol were evaluated by merely asking "was alcohol present?" we would massively overestimate its causal role the same way THC is being overestimated here.
The problem with THC data is not that baseline comparisons are illegitimate; it's that we lack an agreed-upon, time-linked impairment metric comparable to BAC. THC metabolites persist long after intoxication, so presence alone is a weak proxy for risk.
So applying baseline controls to THC is not "apologism", it's applying the same evidentiary standards we already demand for alcohol, so the opposite of what you said.
OpenAI would still lose money if the basic subscriptions were costing $500 and they had the same amount of subscribers as right now. There's not a single model shop who's ever making any money, let alone ungodly amounts.
These costs you are referencing are training/R&D costs. Take those largely away, and you are left with inference costs, which are dirt cheap.
Now you have a world of people who have become accustomed to using AI for tons of different things, and the enshittification starts ramping up, and you find out how much people are willing to pay for their ChatGPT therapist.
.jpg){kind=link}
reply