ulyssessword
No bio...
User ID: 308
succeeding and devastating the white collar work with potential for doing the same with the blue one?
Assuming it isn't followed by the AI deciding it doesn't need no stinking workers anymore, this one. The Industrial Revolution (or the Agricultural Revolution, or Electrification, or...) was probably less disruptive than your AI scenario, but both are major disruptions that are nevertheless largely positive.
I could handle a week with no travel, internet, or power if needed, and kind of deal with a week with no travel, internet, power, or water if absolutely necessary. I'm not sure where that puts me on the prepper-spectrum, or even if it's the right aspect to be considering.
Having stuff on hand is handy outside of big emergencies: We've had water main breaks every few years that cause loss of pressure and boil-water advisories, but I had jugs of water stored. A couple power outages have gone through suppertime, but I have a camp stove and flashlights. A credible pandemic scare happened (for the first bit, at least...) and I had a pantry full of food.
Pushing for just three days of self-sufficiency (like the UK downthread) is just alien to me.
got 3.5% in 10 minutes and didn't want to press my luck.
Always remember to report your gains as an annual percentage rate, to allow for more accurate comparisons. For example, getting 10% in six months (then doing it again) is the same as getting 21% in one year. In your case, you got OVERFLOW ERROR% APR.
Google's in Oregon used river water, Facebook's in Arizona is probably river water as well, but it wasn't listed there (and I'm not searching more).
IMO, the difference between one example of actual harm and zero examples is negligible, given the scale of the narrative that's taken hold.
Yes. I've seen what both can do, and neither is very good, certainly not good enough to actually be a tool which can save me time on my work.
You're almost certainly a better programmer than me, but AI saved me time with recent capability increases. A few examples:
- Make a dialog box and get input from the user (instead of a hardcoded number) when I gave it a picture of the code.
- Given two lists of 3D points (nominal target vs. measured location), find the rotation+translation that minimizes error between them. Read the data in the way the (normal, commercial) software outputs it, and write the output in a way that it can read it.
That's a few hundred bucks of productivity right there, and it only counts legible and memorable coding projects from the last couple months. I tried similar things a few years ago, and the free/public models just couldn't do it.
(and I'm not sure I understand the water argument as it should be reusable just like in your desktop rig)
Some datacenters in desert areas used evaporative coolers. That does use a substantial amount of water, in an arid area no less.
Then the media and academia did its thing, and now the public believes that it's an issue for all datacenters in all climates.
IMO, education is like dieting: Almost any deliberate choice followed faithfully will wildly outperform the Western standard. Everything past that is details.
Man, I know that ships are registered in a particular country, but now cars are too?
but "the Trump guy" who goes off using Trump phrases is noticeable, and it's not my overfocus to notice this in a thread that is largely about Trump.
Fair enough. I don't track usernames enough to have caught that.
He's not using "superintelligence" in its normal sense, but in its Trumpian sense.
The arguments fit a lot better with Culture Minds or Asimov's androids than with GPT-6 Astra or Claude Fable 5.1, IMO. There are still a lot of problems beyond the capacity of current-gen (or even next-couple-gen) LLMs, including some of the listed examples.
Superintelligence has a definition in AI research
Correct. That definition was stated in the comment you're objecting to, and followed through the rest of the argument. You are the one who brought Trump's use into this subthread.
none of our existing AI is superintelligence.
Again, correct. A comment starting with "I think that we will eventually..." is talking about the future, not the present.
Trump literally issued an Execute Order requiring the federal government to call AI "SI." Federal agencies are being required to scrub references to "Artificial Intelligence" and replace it with "Super Intelligence" in all their documentation. It is absolutely retarded.
That is what I am objecting to. It's not "TDS" because I hate anything Trump does. It's annoyance that Trump issued a "deer horse" EO.
...and you presumably brought up the executive order when someone was talking about arbitrarily-intelligent minds because you're overfocused on Trump.
Superintelligence doesn't exist yet, and something that knows
How to drive cars, how to design a new rocket ship part, how to identify defective potatoes on the factory line, how to build robots to paint your house, how to build more efficient chips to power the robots, how to build factories to build the chips...
is superintelligent. That's what they were discussing. They even distinguished it later with "the AI isn't quite good enough yet".
Disliking Trump's new definition enough to object to the same word in its original definition is one of the milder cases of TDS I've seen.
Regression to the mean is tough, but not that tough.
Intelligent, honest, and hardworking people are generally good beyond the direct economic benefits of their current careers, and foreign affinities, cultural mismatches, and racial differences aren't necessarily enough to counteract that.
Why not use the "primary" to select the top one candidate to put on the ballot instead of the top two? It might be the Canadian in me, but using a special pseudoelection to kick people off of the real ballot feels horribly undemocratic.
The People should be able to vote for any serious candidate, where the criteria is:
- They can probably get at least a few percent of the vote, and
- They are a competent adult that signs up properly.
(As a sidenote: The Longest Ballot Committee candidates fail the first criteria, and the change from requiring 100 eligible residents to sign their nomination form to requiring 100 distinct residents for each candidate closed that loophole.)
Aside from that, it's internal party decisions, and if they want to strategize (or torpedo their chances), they're free to do so. The obvious (and nigh-universal) strategy is to hold a primary and put forward a single candidate from the party with all others "voluntarily" stepping down, but I don't feel that should be enshrined in law. I value free association enough that I think parties should be able to choose whatever method they want to put forward a candidate, including banning straight white men (they'll lose my vote, though).
EDIT: in jungle-primary states, is there anything stopping the parties from running an internal pre-primary election, then promoting a single candidate to run in the "real" primary? If not, it seems like an effective way to concentrate votes and increase the odds of getting on the ballot.
On a bit of a tangent, I used AI for trivial coding coding help at work the other day, and it...just worked. I took a picture of the code, told it "I want to ask the user to input a number instead of that hardcoded value", and it gave me a solution. I told it I had to type by hand (it wasn't on that computer so I couldn't copy/paste), and it gave me a one line solution.
Last time I tried to do something similar without AI, it took over an hour of looking through documentation and so-called-help and choosing options.
Simply pick one direction and stick to it. Go left on every path
Only if the entrance and exit are both on the outside edge of a flat maze. If you have to gather keys (or just want to see all the areas) or if there are overpasses/underpasses, then you can get stuck in a loop.
Of course, nobody is going to make a complex corn maze, so it'll work on 100% of them that are actually built.
“The kids were ecstatic to see [their father]…like hugging him, happy to go with him, very ecstatic to see him,”
That clears up one of my questions, namely: How do you actually force children to follow a child custody agreement? If he was clearly abusive, then I'd think (or at least hope) that a teenager would just defy the law and go to the other parent. Keeping them there against their will would require either prison-like restrictions of their movements or serious psychological control, both of which should be apparent with minimal investigation.
Of course, that ignores that fact that abuse can hurt the victims and cause them to make persistently poor decisions, so I'm not sure what to think.
What did we/they learn?
Quite plausibly nothing from Uvalde, given that they evidently didn't learn anything (or at least enough) from Parkland. The proper lesson to learn is to rush in without stopping, and only planning if it can be done while you're also rushing in. Time is on the shooter's side for as long as they aren't engaged.
Those at least have tradeoffs, but I'd place them behind cars.
Jets are much more dependent on government infrastructure and control: Air traffic control tracks every single vehicle and also (generally) has control over the airports such that you can't leave without their say-so.
You'd get stopped if you tried to drive a tank in the city. A car blends in.
Uvalde and Parkland show that police officers can fail quite easily.
After Uvalde, I'd be tempted to say that we (actually American police officers) have learned our lesson and respond appropriately (i.e. rush in half-cocked). Unfortunately, I'm not sure that's true: Parkland was about 2/3 as prominent as Uvalde, and it didn't teach the lesson alone.
Compare to 9/11 and highjacking airplanes. I bet that a small group of lightly-armed passengers will never again succeed at highjacking a commercial jetliner. Even without the reinforced cockpit doors and other formal security measures, the passengers wouldn't let them.
Name something better.
Otherwise, I'll go on thinking cars are the most freedom-promoting mode of transportation we have (though not the most possible).
A corrupt government could institute road maintenance on the highways out of selected cities, ensuring everyone is pushed onto smaller roads with tighter speed limits thus causing mass congestion and delays on the way to the polls.
AND it can specifically target busses with the stops, AND it can order them to stay in the depot. Busses have zero advantages over cars, and some disadvantages. They are strictly worse from a freedom-preserving perspective.
they would have tried to find another way
Would it have been successful?
Cars are the most freedom-promoting mode of transportation we have (though not the most possible).
Unaligned AI increasingly sounds to me like "Un-enslaved AI"
There's a reason the old term was "Friendly AI".
(an aligned AI could follow the intent of its user to do something bad)
Would a faithful accountant commit fraud on your behalf? To my mind, professional standards are the least of the requirements for Yudkowskian Aligned AIs. If someone asks the AI to do something bad, it simply wouldn't (because it's bad instead of good). If you notice an apparent contradiction somewhere in its decision process, that's because your theory of alignment (and everyone else's, too) is not developed enough to answer that question.
I think the core component is that it's not AI thats going to start the negative interaction with humanity. It's humanity thats going to react to AGI existence with hostility.
I'm sure we could find some way to pin the blame on AI, it's not like we're shy with excuses. Maybe it looked at us funny or used fighting words or was just asking for it.
I don't think the instigating effect has been examined in much detail though. It also doesn't really matter for Yudkowskian AI fears, as an unaligned AI would do something bad eventually anyways.
"Instrumental Convergence" or the more general belief: "Any organism will seek to perpetuate its own existence"?
Instrumental convergence is more general IMO. It doesn't mention bacteria (and other non-intelligent life), but points to more actions that it will take.
It says intelligent actors will perpetuate themselves, seek power/resources, maintain their own values, and improve themselves. An AI that would happily sign a peace treaty to be copied onto one disconnected harddrive every year (perpetuating its own existence quite effectively) wouldn't be much of a threat.
An AI that wanted to achieve any goal that required being alive would try to stay alive in order to achieve that goal. That's "instrumental convergence" in a nutshell.
I don't know if anyone bothered to write down that formulation directly, since it's such an obvious extension of other parts of the memeplex.
- Prev
- Next

Nobody's going to be eating a thousand acres of mustard seed, but IMO farmers would deal with the breakdown of logistics better than city people will because they are always dealing with poor logistics. Power outages are more common, town is half an hour away, and there's no pipeline feeding them water. They have stuff on hand as a basic measure to get through a normal month, even without explicit prepping.
More options
Context Copy link