ben_w

8 ms·

61.
by ben_w 25d ago
More like: Your Honour, it is indeed correct that during testing of our new missile to see if it could correctly function in the military role of seeking out and then breaking through a blast door, the targeting system locked onto the n
62.
by ben_w 25d ago
The hacking should definitely be treated seriously. It is a (fortunately low-stakes) instance of the general problem, and we are lucky (at least, we think we're lucky!) that the tests didn't include a similarly poorly phrased requ
63.
by ben_w 25d ago
> If we passed a law that said the CEO of any company that deploys an LLM that commits a crime gets punished as if they personally did the crime (so, basically instant life sentence if it's even a simple crime times a million instan
64.
by ben_w 25d ago
> Start making managers pay the price for their actions, and watch how the models magically slow down on their own. The problem here can be personified as "Trump", and it's the same problem that applies to coal. Coal has a
65.
by ben_w 25d ago
If my local government or some local business was limited to using one specific newspaper, this would indeed be a problem. And in reverse, having a concentration of readers (users) on one platform, means that whoever controls that newspap
66.
by ben_w 25d ago
> If you don't like the content on X enough to make an account The content itself isn't the problem, it's the website. Other people keep linking me to X posts they want to share. The website is so bad that it (was) faster
67.
by ben_w 25d ago
Does the market crave good AI experiences? LLMs are eating software, maths, and LinkedIn posts; some of the image generators are seeing similar (limited) success. Other than that, my general impression is that most people think of even the
68.
by ben_w 26d ago
Reminder: it's not the offline un-prompted weights sitting alone on a hard drive that anyone is worried about. It's what they do when prompted (both when it is and when it isn't what the user asked for), that we get the most
69.
by ben_w 26d ago
Would you say merely "good", or is this agreement on "fantastic"?
70.
by ben_w 26d ago
The other half of "cost" is "return on investment". If I had a plan to e.g. copy-paste specific success stories from any rich country to any poor country, something that might e.g. raise the per-capita income of the poor
71.
by ben_w 26d ago
Everywhere else but the USA (and possibly Luxembourg), $90k will buy you a fantastic quality of life. If you really want to make $90k seem as bad to the rest of us as it feels to you, you need to get the exchange rates changed, have USD dev
72.
by ben_w 26d ago
Rest of us really ought to put tariffs on US software. Not to bring in jobs or money (tariffs don't do that); just to push us away from your companies.
73.
by ben_w 26d ago
Someone recently rage quit one of the linux distros because the maintainers decided to accept LLM generated code (if it was good). Other distros ban use of LLMs. Not sure how that's going to go down given Linus seems to be willing to u
74.
by ben_w 26d ago
My take from the failure of JTB is that "knowledge" is an illusion. Best we can do is have a belief, any belief will unavoidably be justified, but we have no access to any oracle which can tell us if any particular belief was ever
75.
by ben_w 26d ago
Because nobody knows what they're doing, everything is a heuristic learned by another (simpler) AI whose job is to guess how humans would rate responses. There are regular attempts to force regularity between human categories, have the
76.
by ben_w 26d ago
The pets seem to be well cared for, and given a nicely enriched life with plenty of metophorical chew toys and laser pointer dots to chase.
77.
by ben_w 26d ago
Exponential growth curve(s) has yet to flatten; but you do also need to account for capacity factor, which is different between wind and PV.
78.
by ben_w 27d ago
Non-AI-assets: yup, this is sensible. AI is a signal of being unconcerned about being seen as a cheat, absence is good. Non-AI-code: forget it, you won't be able to avoid it in your libraries, and you won't write the whole engine
79.
by ben_w 27d ago
Renewables growth is so fast that it will exceed 100% of electricity in the early, and 100% of all form power in the late, 2030s. It's the meat methane and cement CO2 that's now a big question.
80.
by ben_w 27d ago
I have good news and bad news. The good news is biologists seem to think that's fairly difficult, so much so it will (probably) not be a nearterm threat. The bad news is: there are definitely people who want that goal, and the AI is cl
81.
by ben_w 27d ago
The people who think as you think indeed have left or never joined. The people still there necessarily think they can make a difference. I never applied to any of them because I didn't think I could make a difference. I currently have
82.
by ben_w 27d ago
> not the AI going "out of control" or whatever and killing the human race All of the "going out of control" examples can be described as examples of "human asked computer to do thing". The only question is
83.
by ben_w 27d ago
> Do you think we should treat them as liars or threats? I answered that with the analogy you called "spin" and "avoiding the question", and completely misunderstood because "vague insinuation of climate change d
84.
by ben_w 27d ago
Dunno about "should"; personally I am more scared of the US than China here, but I do know how propaganda can be, makes "should" really difficult. > Will OpenAI and Anthropic open themselves for inspection so we can v
85.
by ben_w 27d ago
You're not even close to passing the ideological Turing test here.
86.
by ben_w 27d ago
Rather the opposite.
87.
by ben_w 27d ago
> Serious response: we have no idea what ASI means or might be like. It's literally sci fi (at the moment). Yes. Unfortunately, this is part of the problem.
88.
by ben_w 27d ago
Today? As in, you don't foresee any new models or categories? Fine. Providing sychophantic assistance to multiple mutually-incompatible genocidal racial supremicists.
89.
by ben_w 27d ago
I say same about p(doom): Only makes sense in a time interval, otherwise you have to ask how people categorise a hypothetical AI which fails to overcome the heat death of the universe. (When I say mine is 0.1, this is "in my lifetime a
90.
by ben_w 27d ago
One of Europe's current (multi-)national security concerns is how much of its various economies are dependent on US software, cloud, and that a Trump executive order sanctioning any subset of those countries may terminate those economi

More ›