Very interesting - why aren’t you able to access them outside of a web API right now?
Have you tried smaller quantized open weight models that aren’t frontier? They probably can’t automate all your coding but I imagine they could at least help a lot with the drudge work that just takes a lot of time but isn’t necessarily complex?
Yeah that’s interesting - I feel like smaller chunks of work that aren’t inherently complex like class definitions and utils etc are actually one of the nicest things LLMs help automate - I can understand a human modifying the output directly to finesse it to exactly what they want but I’d be curious why someone wouldn’t want to use an LLM to generate all the boilerplate and make a quick first draft of something that gets 80-90% of the way there for these kinds of things.
Definitely curious if anyone is actually doing this fully by hand and if so, why!
Surprised I completely missed this book coming out last year and that I haven’t seen it on Hacker News despite checking this place constantly.
Found this to be the most compelling and best written AI doomer take, and I’ve yet to find a strong counterargument. Posting this in part to source counterarguments. Anyone have any good ones?
When you open up Activity Monitor, to the immediate left of the "Memory Used" and "Cached Files" that you see, you'll see the Memory Pressure graph that the guy above is talking about.
On my 64 GB M1 Macbook Pro right now, I have 53.41 GB of Memory Used and 10.72 GB of Cached Files and 6.08 GB of swap, but Memory Pressure is green and extremely low. On my 8 GB M1 Macbook Air I just bought for OpenClaw, I'm at 6.94 GB Memory Used and 1.01 GB of Cached Files with 2.05 GB of Swap Used, and Memory Pressure is medium high at yellow, probably somewhere around 60-70%.
You can open up the Terminal and run the command memory_pressure to get much more detailed data on what goes into calculating memory pressure - more than just the amount of swap used, it tracks swap I/O and a bunch of page and compressor data to get a more holistic sense of what's going on and how memory starved you're going to feel in practice.
In any case - I've been absolutely mindblown at how fast my 3 8GB M1 Macbook Airs I just bought for ~$350 brand new have been - even with tons of Chrome tabs open, multiple terminal windows open, running OpenClaw and Claude Code and VS Code and doing a ton of development and testing, never once have they ever felt slow. Oftentimes they actually feel faster than my 64 GB M1 Macbook Pro, which kind of blows my mind and makes me wonder wtf is going on on my monster machine. Moreover, my M1 Macbook Pro drains battery like crazy and uses a ton of charge, whereas the Macbook Airs stay constantly below 10 watts essentially always and even with Amphetamine keeping them on 24/7, with the display off and being fully on, they'll drop to a single watt of power draw. Truly insane stuff. I've lost all my concern about RAM, to be honest (which is shocking coming from someone who bought a top of the line maxed out RAM primary machine in 2021 specifically because I felt like RAM was so important)
Masking is a learned behavior - learned specifically because the person in question had sufficiently unpleasant experiences being unmasked that they decided they needed to try hard to mask in order to avoid said unpleasant experience.
No one masks at birth. If you want your child to stop masking, perhaps it’s helpful to investigate what caused them to learn the behavior in the first place and what could be done to make the experience of being unmasked more pleasant and less aversive for him.
Just as you say, masking takes an incredible amount of energy and is exhausting and often backfires. Why would anyone expend such exhausting amounts of energy without some extremely strong motivating factor? The alternative to not masking must be perceived as exquisitely undesirable.
The truly amazing thing here is how powerful the placebo control was, even despite the fact that as someone who's done ketamine IV treatment before, it's pretty obvious when you're on ketamine and when you're not.
"At week 6, remission in the ketamine arm remained high, although non-significantly versus placebo (69.5% v 56.3%; odds ratio 0.8 (95% confidence interval 0.3 to 2.5), P=0.7)."
^^ Eating animals. Firmly convinced we're on the wrong side of history on this one - when the world finally stops doing it en mass, either because clean lab grown meat has become economically more viable or environmental concerns are too overwhelming, etc., very convinced we'll look back on it as a horrific practice just as we might slavery today.
Aside: this thread escalated quickly from scamming Apple gift cards on reddit
"Just like us humans, learning to drive and navigate seemed to have a relaxing effect on the rats. In a control experiment, they found rats had higher levels of cortisol when being driven around in remote-controlled cars than when they were allowed to steer themselves."
Seems like a terrible control imo for me to say driving intrinsically has a relaxing effect. A better control would be rats doing nothing vs rats driving and seeing if the rats doing nothing were more stressed than the driving rats, meaning the driving rats were actually lowering their baseline stress.
In this experiment the control of being a passenger in a terrifying vehicle moving all by itself with no autonomy or control over the situation can quite likely be the thing causing elevated stress, rather than rats driving depressing levels of stress. Imagine if someone suddenly strapped you into a bubble that started moving by itself and you have no idea why this is happening, no control over the situation, or where you're going or what's going to happen to you. Stressful af. Hell, people get stressed just being in the passenger seat watching someone else drive.
The vast majority of people prefer having autonomy and control over their own motion vs being helplessly navigated by someone else you don't know/trust with zero context and no idea what's going on. A little misleading if this reflects the actual study.
The article directly addresses the idea that "Being 17...doesn't make you less culpable."
"Of just over 200 released, six have faced new charges, but only one has been convicted of a new crime, contempt. That’s less than a 3 percent recidivism rate. “When you’re looking at the percentages of the guys who’ve come back in court, and compare that to our overall, which is around 40 percent, you really know you’ve done a good job,” Johnson said.
That outcome supports a theory the U.S. Supreme Court first advanced in 2005, when it outlawed the death penalty for juveniles. The Court took stock of “evolving standards of decency,” as well as a growing body of neuroscience suggesting kids’ immaturity makes them less culpable than adults. It also emphasized that kids, more than adults, are amenable to change. The low recidivism rates likely also reflect the reality that criminal behavior declines dramatically as people age out of their teens and early 20s."
There are a million things incredibly unwise things I would have been far more likely to do at 17 than 27 now - I imagine this is a trend that will continue to likely progress for me individually, and I also imagine this isn't a phenomenon singular to myself.
The shockingly low recidivism rate seems to empirically support this notion as well, particularly when contrasted with the general recidivism rate. If these juvenile offenders were just as culpable as anyone else when they had committed their life sentence crimes, presumably their recidivism rate should be similar. 40 percent for the general population and 3 percent for this cohort so far suggests otherwise.
As someone who went to Harvard, I'm pretty sure there are a lot of extremely decent universities that all basically provide the same quality of instruction. People just get unreasonably hung up over the 'best' schools, but there are easily at least like 100+, if not several hundred, that provide easily roughly the same educational experience as far as I'm aware.
The peer group may definitely dramatically different from what I can tell, but if you're a serious student and there for the learning above all, very straightforward to get into a school that will teach you just about anything you could learn at Harvard or any other Ivy League or similarly 'elite' school.
Fairly shocking to me that Yuan Cao received literally a single glancing mention here, but he was clearly the one who actually made the pivotal discovery, and is the first author on the Nature paper (https://www.nature.com/articles/nature26154), and was the one named to the Nature 10 in 2018 for his discovery.
Really mind boggling that this entire piece focuses all the attention on his supervisor and almost neglects to mention him at all.
Have you tried smaller quantized open weight models that aren’t frontier? They probably can’t automate all your coding but I imagine they could at least help a lot with the drudge work that just takes a lot of time but isn’t necessarily complex?