AGI can’t be defined as autocomplete with fact checker and instinct to survive, there’s so so so much more hidden in that “subsystems point”. At least if we go by Bostroms definition…
If you get off the internet you'd not even realise these tools exists though. And for the statement that all jobs will be modelled to be true, it'd have to be impacting the real world.
Is it even possible to "get off the internet" without also leaving civilisation in general at this point?
> it'd have to be impacting the real world
By writing business plans? Getting lawyers punished because they didn't realise that "passes bar exam" isn't the same as "can be relied on for citations"? By defrauding people with synthesised conversations using stolen voices? By automating and personalising propaganda?
Or does it only count when it's guiding a robot that's not merely a tech demo?
I’ll be worried about jobs being removed entirely by LLMs when I see something outside of the tech bubble genuinely having been removed by one - has there been any real cases of this? It seems like hyperbole. Most people in the world don’t even know this exists. Comparing it to the internet is insane, based off of its status as a highly advanced auto complete.
Sure, but think about all of the jobs that won't exist because this studio isn't being expanded, well beyond just whatever shows stop being produced. Construction, manufacturing, etc.
Edit: Also this doesn't mean less medea, just less actual humans getting paid to make medea or work adjacent jobs
Maybe it's time to construct some (high[er] density) housing where people want to live? No? Okay, then maybe next decade ... but then let's construct transport for them so they can get to work, how about some new subway lines? Ah, okay, not that either.
Then I guess the only thing remains to construct is all the factories that will be built as companies decouple from China.
> Comparing it to the internet is insane, based off of its status as a highly advanced auto complete.
(1) I was quoting you.
(2) Don't you get some cognitive dissonance dismissing it in those terms, at this point?
"Fancy auto complete" was valid for half the models before InstructGPT, as that's all the early models were even trying to be… but now? The phrase doesn't fit so well when it's multimodal and can describe what it's seeing or hearing and create new images and respond with speech, all as a single unified model, any more than dismissing a bee brain as "just chemistry" or a human as "just an animal".
Sure and there’s endless AI generated blog spam from “journalists” saying LLMs are amazing and they’re going to replace our jobs etc… but get away from the tech bubble and you’ll see we’re so far away from that. Full self driving when? Autonomous house keepers when? Even self checkout still has to have human help most of the time and didn’t reduce jobs much. Call me a skeptic but HN is way too optimistic about this stuff.
Replacing all jobs except LLM developers? I’ll tell my hairdresser
Right, that entire internet think was complete hype, didn't go anywhere. BTW, can you fax me the menu for today?
And that motorized auto transport, it never went anywhere, it required roads. I mean, who would ever think we'd cover a huge portion of our land in these straight lines. Now, don't mind me, I'm going to go saddle up the horse and hope I don't catch dysentery on the way into town.
I don't think anybody's denying that revolutions happen. It's just that the number of technologies that actually turned out to be revolutionary are dwarfed by the number of things that looked revolutionary and then weren't. Remember when every television was definitely going to be using glasses-free 3D? People have actually built flying cars and robot butlers, yet the Jetsons is still largely wishful thinking. The Kinect actually shipped, yet today we play games mostly with handheld controllers. AI probably has at least some substance, but there's a non-zero amount of hype too. I don't think either extreme of outcome is a foregone conclusion.
Capabilities aren't the problem, cultural adoption is. Just yesterday I talked to someone who still googles solutions to their Excel table woes. Didn't they know of Copilot?
Maybe they didn't know, maybe none of their colleagues used it, their company didn't pay for it, or maybe all they need is an Excel update.
But I am confident that using Copilot would be faster than clicking through the sludge that are Microsoft Office help pages (third party or not.)
So I think it is correct to fear capabilities, even if the real world impace is still missing. When you invent an airplane, there won't be an airstrip to land on yet. Is it useless, won't it change anything?
SICP doesn't teach assignment till chapter 3 if I recall correctly, and that was used for years at MIT. I don't think it set CS back there, so maybe it was more a teaching issue?
> The main use of the Minovsky particle was in combat and communication. When the Minovsky particle is spread in large numbers in the open air or in open space, the particles disrupt low-frequency electromagnetic radiation, such as microwaves and radio waves. The Minovsky particle also interferes with the operations of electronic circuitry and destroys unprotected circuits due to the particles' high electrical charge which act like a continuous electromagnetic pulse on metal objects. Because of the way Minovsky particles react with other types of radiation, radar systems and long-range wireless communication systems become useless, infra-red signals are diffracted and their accuracy decreases, and visible light is fogged. This became known as the "Minovsky Effect".
The disruption of electromagnetic radiation is due to the small lattice of the I-field creating fringes that long wavelengths cannot penetrate, and that diffract wavelengths that have similar distance with the fringes. This diffraction and polarization process disrupts the electromagnetic waves.[23]
The only counter measure to the "M" particle in the series was to install bulky and expensive shielding on all electronic equipment, but only to counteract the effect it had on electronic circuitry. While this could be done for space ships and naval ships, this ruled out the use of precision guided weapons, such as guided missiles. Due to this, the military use of Minovsky particles ushered in a new era of close-range combat. This is the primary reason for the birth of the Zeon close-combat weapon: the mobile suit.[24]
"There was a time when battles could be won and lost with the mere push of a button, and these mobile dolls are the absolute root of that detestable and hateful spirit. When war is dehumanized both victory and defeat become miserable, and God no longer lends a helping hand." - Treize Khushrenada
Not UC, I know, but it's the first thing that came to mind when I saw this thread.
The reviews on the app are wild - people talking about loving getting paid a couple bucks for watching it all day, and that it's their new side gig. It's pretty sad, to be honest. Get outside and live life! (Or stay inside and live life, but we need to get away from this dopamine high of doom scrolling)
I see a lot of fuzzing tools for CLI apps, but are there any good alternatives for web applications/APIs? I've used Hypothesis for generating random datas in requests but maybe there's something better out there.
Heroin is less social than alcohol. You don't go on a date and take heroin. Because the effects don't match what you want. To be fair you don't go on a date and take cocaine really either, because you "won't work". Parties with only cocaine or heroin would suck.
But yes they are more dangerous to the individual and would be more dangerous to society if everyone decided to start overdosing heroin all the time! But they'd all die and then alcohol would be back at number one.
> Well, obviously, far more people drink alcohol than use those two.
But saying alcohol is the most dangerous drug with this definition is still fair enough to say right? Carfentanil is more potent than fentanyl, but fentanyl is on the streets so it's what we should crack down on.
Per capita, alcohol kills more people than any other drug in most countries. It's also one of two drugs where you can die from withdrawals, being more dangerous than crack or heroin to detox from. It's also the only drug you get advertised 24/7, and that is socially forced down your throat (literally).
Not to say alcohol isn't fun, but it is the most dangerous drug to society, statistically.
>But saying alcohol is the most dangerous drug with this definition is still fair enough to say right?
This is conflating the difference between "danger" and "harm". Danger is a measure of risk, not damage.
You wouldnt say that drinking alcohol is more dangerous than drinking bleach or sulfuric acid, but it can be more harmful, if you are talking about rates.
Is this a good metric? For a potential drug consumer it is more relevant which health/social risk is involved if taking a certain drug once/sometimes/regularly.
Alcohol causes 5.3% of deaths globally. While at an individual level alcohol is only unhealthy on average, as a society its impact is pretty significant. While some drugs are definitely worse at an individual level, access and criminality should mean they're less of an issue. If we're going by purely individual risk not taking into account other factors then 14-Methoxymetopon is 10,000 times stronger than fentanyl.
The issue is the misleading presentation of it. The writer of the linked blogpost is either dishonest or already missing what you're saying.
>Next time someone tries to justify drinking alcohol, remember it causes cancer and other diseases, it’s three times as harmful as cocaine or tobacco
I know people don't understand cumulative effect, and you know they don't understand per capita. It's just a dishonest way to present something. You could say it's fair that driving is more harmful than drunk driving, but that'd be misleading at best.
Sure you're right, but the conversation that heroin is more likely to kill you than alcohol is (or should be) obvious. However people don't seem to realise 3x more Americans die per year from alcohol overdoses than opioid overdoes or that it is the cause of 5.3% of deaths globally per year. Parents rarely buy their developing kids heroin to take to a party, and you rarely go out after work with your boss to see who can do the more lines of cocaine.
You're more likely to know someone who dies from alcohol than any other drug, if you don't already. And I say that as someone who lost a friend to heroin! So I get how bad it can be, and I get how fun alcohol can be, and how culturally accepted one is over the other, but damn alcohol is a deadly drug. Like I said in another comment, alcohol is one of two drugs you can die from withdrawals from too (the other being benzos, ironically what you get to help deal with alcohol withdrawal).
And I say this knowing I still fancy a beer after work! Because that's how ingrained this stuff is in our lives.
I would argue soda is a lot more impactful than alcohol if we talking about cumulative deaths, financial cost, or quality adjusted lives lost.
The reality is that there is many ways to compare the characteristics of different substances. There are long-term impacts, short-term impacts, ld50, moral impacts, dose control, chemical addiction potential, chemical addiction rate, Etc.
You have to understand all of these to make sense of why different chemicals might be considered differently.
Like it or not, When compared to a lot of drugs alcohol has a high ld50, High dose control, and low chemical addiction rate.
You are unlikely to OD on a drink because the dose was high and there is too much alcohol in it. You are unlikely to get chemical withdrawal and cravings from infrequent use.
This puts a lot more control in the hands of users which is an important moral and Regulatory consideration.
Utilitarian analysis of total death or cost to a very bad job of taking into account any of the moral considerations
