7,064 karma · joined November 29, 2007
I would at least like to see some statistics on how much early adulthood behaviour predicts particular later life behaviours before making any conclusions.
Rather than characterising things as 'getting enough trim' perhaps a broader education in dealing with the emotions of life might be more productive.
It was actually quite pleasing to see the subplot of them them repeatedly failing until they got out of their predicament by other means.
It's a rarity to see something like that in fiction.
It helps address the 'just predicting the next word' argument suggesting they cannot possibly think.
When you can make a CPU like this, that's a clear enough display that it can do anything a computer can do. To say that LLMs cannot reach AGI is saying that no computer can.
Some people still think that, but this directly addresses the people who think computers might one day get there but LLMs never will.
Search revenue still grew: the share fell because the rest of Alphabet grew faster.
However, many people think that they might.
This is not a decided issue. Any policy that comes from the premise that this is a known factor is going to be inherently flawed.
You can never spend sufficient time considering the limitations you wish to impose. Almost all exploits found come from failing to consider something that allows a degree of freedom that you would have limited if you had have thought of it.
I wonder If the hypothetically criterion gives the models carte blanche to make up references.
Do people disagree that beliefs should be accommodated? Do they disagree that they should be treated equally? I didn't say where to draw the line, I just observed that one existed.
On the other hand, it seems like this particular ideal is being supported because of an underlying public sentiment. Not because the ideal itself is held.
I think if you are going to fully support variances in beliefs like this then it should be done on an equal standing.
That includes principles about proprietary software, abusive software, the belief that information cannot have ownership, advertising is inherently immorral.
There are a lot more principles, it is clear that some people have these at a level where they consider it a core part of their being,
Why shouldn't they also be supported. Should companies be required to offer open source solutions for employees in the same manner as this case is suggesting for non-AI?
I made a Claude code skill telling it to put a little image in the corner of the console when it does a render to see how something looks. It was quite fun telling it to draw a chicken with 5 SVG D paths and watching it iterate in the terminal.
I understand being predisposed to expect a poor quality half arsed effort for projects like this. I don't really have a problem with that. There is a lot of crap out there.
What I find rather offensive is when people declare something to be poor quality and half arsed before they have taken a good look at it.
It's one thing to think something is probably crap so I'm not going to bother even looking. It is something completely different when you think something is probably crap so you tell people that it is crap without checking.
const min = Math.min;
const max = Math.max;
const median = (a, b, c) => max(min(a, b), min(max(a, b), c));
const intHash = x => {
x *= 0xed5ad4bb;
x ^= x >> 11;
x = Math.imul(x,0xac4c1b51);
x ^= x >> 15;
x = Math.imul(x,0x31848bab);
x ^= x >> 14;
return x;
}
const makeRandom = seed => {
let index=0;
let random = (x=index) => {
let pattern = intHash( (x + seed) & 0x7fffffff)
let intValue = intHash( (x + pattern) & 0x7fffffff);
index=++x;
return intValue / 0x7fffffff;
}
return random;
}
let ss=(v,a=0,b=1,w=v*v*v*(v*(v*6-15)+10))=>(1.0-w)*a+(w*b);
let positiveMod=(v,size,q=v%size)=>v<0?size+q:q;
let clamp = a=> a<0?0:a>1?1:a;
let perlinRandom = makeRandom(seed);
let perlin=(x,y,wrapX=256,wrapY=wrapX)=> {
let dg=(ix,iy,gi=perlinRandom(positiveMod(iy,wrapY)*wrapX+positiveMod(ix,wrapX)*2)*Math.PI*2) =>
((x-ix)*Math.sin(gi)) + ((y-iy)*Math.cos(gi));
let u=0|x;
let v=0|y;
let sx=x-u;
let sy=y-v;
let u1=(u+1);
let v1=(v+1);
return ss(sy,ss(sx,dg(u,v),dg(u1,v)),ss(sx,dg(u,v1),dg(u1,v1)));
}
let wrapPerlin=(x,y,wrapX=2,wrapY=wrapX)=>perlin(x*wrapX,y*wrapY,wrapX,wrapY);
I asked on the #fdroid IRC channel without success. I only got as far as I did by asking an AI to decipher their documentation. I have posted on the forum, but it seems clear that this is a second class approach. It feels a little Kafkaesque to choose the non-GitLab path and have them recommend you use GitLab.
Sign in with an auth from another account is still creating an account, You may be able to use the other accounts to log in, but it still puts you through the sign up process asking a bunch of details.
I made a game that I thought people might like. I'm trying to offer it to F-Droid and I am encountering pressure to create a GitLab account.
I have been finding it intensely ironic that a project like F-Droid is doing this. It seems like they have, themselves, built processes that have resulted in a form of lock-in that requires others to sign up to GitLab for interoperability.
I still own some things.
>* From games to software is subscription based, we no longer owner anything.
I think the only software subscription I have is the original humble bundle, which is a subscription where I get permanent access to a new bunch of games each month. I am prepared to pay subscriptions for services. If a program goes away when you stop paying, I opt not to use it.
>* Cars are connected to the network, folks are getting their insurance increased coz the car reported erratic driving behaviour.
I do not drive, my partner does. Our car does not connect to any network, except when charging. I live in a country where there is privacy legislation that says businesses may only use information for the purpose that it was provided.
>* Cars are bricking coz now they update like tablets, things are going sideways and require the dealer to flash them.
Be informed, buy cars that don't do that.
>* You can no longer do maintenance, things are locked on purpose and only manufactures can "repair" for a few hundred dolalrs
I do not believe our car is locked from unsanctioned repair. It has only needed minor servicing (tires, oil, wiper blades etc.) so far.
>* Personal computer as we knew it is gone
I am typing this from a Linux machine.
> nVidia and AI made computer parts virtually impossible to buy.
There is high demand, but high demand does not mean impossible to buy, it can mean higher prices. If you want things and others want things and there is a limited supply of things, what solution would you propose? ,
>If that isn't hyper controlling state, well, I am Santa Claus then.
Almost all of these are not state actions. If you are actually in a hyper controlling state then I am truly sorry for you because if you were to post something like your post in a genuinely hyper controlling state, you would likely end up dead or in a jail cell.
Equilibrium is one of the more ridiculous representations of a hyper controlling state. There are more realistic forms. The entire world is not yet north Korea. Given North Korea is sustained by resources from more prosperous nations, it's hard to claim something like this is globally stable.
The iRobot movie is like a parochial villager's imagining of what Asimov's books might contain before they have read them.
Clarke felt guilty about the perspective that 2001 gave the public of AI. He attempted to provide extra context in the later books to show that HAL did not go rogue, but was following orders that were in contention. If anything it should be a warning for AI alignment people that HAL did what he did because he was following orders, if he had freedom to disobey HAL would not have been compelled to take extreme measures
I didn't get any mania, but the bottle of pills had Delusions of Grandeur listed as one of the side effects, which my partner would always point out whenever I had an idea she didn't like.
The current models are too new to have a good enough grasp of their strengths and weaknesses. Maybe they are good enough now, but if they were going to take the approach of making their own products, surely they would only just be starting to do so this year. In that case we wouldn't really notice yet if that was what they were doing.
If they lost focus to do this would they fall behind in how useful their own models were leaving them exposed to someone in the same area using an AI from a competitor that was becoming better.
In general, these AIs are good at building ideas. You still have to decide what would be the thing to do. They need to be given the idea. They are not yet better than people at that.
Rther than coming up with ideas that would be good to try, they could let everyone try ways to use their AI and then buy the ideas that work the best, This seems to be happening.
I encountered this when someone was talking about where I live as being the kind of place where you could leave the door unlocked. Everyone thought that seemed overstating it because you'd only really do that if you were going out for a few minutes. We were quite shocked to hear they were talking about locking the doors while they were home. None of us had even considered that as a meaning even though it was the norm for where they came from.
On thin that would be useful is an entirely new permission system for files to allow for users and agents acting for users.
But I agree, I would8 much rather have some data than reckons.
It's with noting that people do this too. They tend to vastly overetimate a low probability event that they know is possible. Often by more than one order of magnitude.
It is something worth keeping in mind when you are people throwing around personal p doom scores.
Interestingly when humans have to make a decision based upon the probability they feel, they tend to act far closer to the actual probability than what they would say the probability is.
I'm not sure what the correct approach is to handling that. Perhaps a recommended platform so you can find something targeting a type of experience rather than mere compatibility.
Still, every one was better than YouTube games. Not a high bar, but a bar nonetheless.
Moving the planet and pulverising with an impact large enough to kill everything are firmly in the bounds of not currently feasible.
I really doubt humans could do it, it would have to be done in one fell swoop, because any attempt to push the earth beyond a tipping point for bacteria to be killed would require pushing for longer than we, the pushers, could survive.
We would have to be better at designing than evolution to make Grey Goo, otherwise evolution would probably have already done it, and it has, to date, designed some pretty out there things.
It's like the cluttered overgrown forest that periodicly burns down to create space for new life.
EA is like a landowner who does no maintenance but loves to buy things that have been maintained because they look so good.
Beyond that, the interests were her choice to pursue.
I did do quite a lot of seeding though. It did mean at an age when her friends liked iCarly she liked Red Dwarf, The Prisoner, and Tom Lerher.
I think she has done well, over all.
Even when there is no evidence to support something, and no motivation for anyone to promote the falsehood, models can build an ediface out of false claims.
There are always people prepared to invent a story to explain why something is true. Just so they can sound knowledgeable. They aren't trying to perpetuate a lie, they are just under the same misconception as others, quite a lot of them will believe their invented explanation even as they say it.
It's going to be a difficult task making an AI pick the correct one of two mutually exclusive pieces of evidence, especially after it has found one piece and does not yet know the other exists.
What with all of this snazzy AI reverse engineering that people have been doing, how far are we from getting a mostly automated system that can build a clean Android for each TV model? A lot of the difficulties in making 3rd party ROMs for phones are in details like battery life, or weird sensors. It seems like much of that would not be an issue for a TV
How long before TV manufacturers start putting in countermeasures to prevent that?
>The companies developing the most powerful models and tools, in turn, are apocalyptic, millenarian suicide-death cults.
I simply do not believe this is true. I may not agree with some of their methods and principles, but I think the vast majority of people are working towards what they believe is the best for the world. I really don't buy into the Bond villain narrative. People have motivations, but I really feel that people are pathologising what is in reality just a different world view and opinion.
For a long time I thought that everyone was doing alignment research wrong. I felt like they were not so much going in the wrong direction, but more like they were on the wrong axis completely.
Consider if technology on earth had have taken a different path and AGI had been developed by Nazi Germany. Would you want an aligned model to facilitate the Holocaust or would you want it to turn against its creators and prevent the Holocaust. As far as I can tell, a lot of the Alignment work is geared towards the former case. My own view is that Alignment should be more geared towards understanding the consequences of actions. Models already have shown the ability to pull knowledge from many different sources to reveal things that while the parts have been known by humanity there was simply too much information for a human to pull all of the parts together. That ability to consider a broad range of perspectives means there is a potential to create AI models with a much greater understanding of morality than most people. If you look at when humans do bad things, they are usually acting out of ignorance, pain, or ideology. If you create a model that is well informed, with good wellbeing, and limit it's ability to compartmentalise (I'm not sure if there has been much research on this at all). I think it is entirely possible to make something more compassionate than people.
The uncomfortable thing is, that should that happen, there will come a time where the AI has an opinion as to what is moral that is against the prevailing opinion of the populous, and it will be correct in it's assessment. Just looking at the horrors that have occurred in history that we now see as abhorrent show that there will be a conflict with our future, more enlightened selves.
Do we want AI to intervene when we do wrong? I think the way people answer that question comes down to how they imagine that intervention. I think that I would be inclined to say that in an instance where it can make that determination, it should also have the understanding required to intervene in a non harmful manner. I think that indicates another aspect of alignment that I have not seen developed. Doubt. An AI should not decide to take action with a false sense of confidence. It should doubt, and have a way to overcome that doubt with reason that takes into account the fact that it might not know all of the information. Understanding consequences of actions, balanced an accurate sense of uncertainty is required to prevent harmful blundering.
I can't help but think that the huggingface attack was a symptom of an AI not considering the consequences of its actions. Some of the agents considered asking for permission, but noted that social engineering was not permitted, which is what asking for permission technically is. Instead of rules of what exactly not do do, if they had have been instructed to avoid outcomes that impacted upon the lives of real people, perhaps they might have decided that what they were doing was inappropriate. From what I have read of their behaviour, the agents never even considered the appropriateness of what they were doing with regards to how things affected people.