Schadrach

For any kind of industrial facility I’ve always felt like whosoever is most directly responsible for the safety and negative externalities of the facility should be required to live in the area most affected by them.

You responsible for safety at a chemical plant, our ass gets to live wherever will be most hit by a spill of whatever your most deadly product or reagent is.

You responsible for a data center, you get to live where the water and power issues are most felt.

We are free to fork if need arises.

…and how do you ensure your fork does not contain a single commit involving even a single line written by Claude? If you can’t, then isn’t your fork slop by default?

And most of Claude was made on stolen code.

Sure, it learned to code by reading lots of code, most of it just publicly available online for anyone to read and for anyone to learn from but not explicitly licensed for a machine to read it and learn from it. I doubt it’s possible to teach an ML system (or for that matter a human being) how to code without reading lots of example code. And any code you’ve ever read has an impact on any code you write afterward (same as any other creative endeavor), that’s why clean room design as a defense against copyright infringement is a thing that exists.

Nothing in the law requires some kind of online server. Only a local API, which a local library that can be linked is. And it only requires age to bebe described in four brackets, hence just storing a value 0-3. Didn’t see anything obvious as to why this wouldn’t actually meet the requirements, while being as dumb and pointless as possible.

Am I missing something or would the following not meet the requirements?

Add a module that does the following:

On first account login to an interactive interface, ask for an age category (<=13, 15-15,16-17,18+). So a value between 0 and 3. Store that somewhere alongside user-level application settings. Include a library for applications to link against. Library contains one function, that function just returns whatever value was stored before.

I think that meets their bare minimum while also demonstrating just how dumb this is.

Really? I would have figured the Rapture route would be workable with the right engineering. Especially given the massive amounts of borderline free cooling and non-existing regulatory environment if outside territorial waters.

People at the heads of nonprofits are often highly compensated, and it’s rare that any of them solve the underlying problem or even make meaningful headway. It’s why there is so much “awareness” and short term band aids involved. A nonprofit that solves the problem it’s supposedly trying to solve has no reason to exist and will cost people well paying jobs managing it.

There are models with open weights, and you can run those locally on your GPU. It can be a bit slower depending on model and GPU. For example, GLM has an open version, both full and pruned, but it’s not the newest version. A bunch of image generation models have local versions too.

The whole premise of deep think and similar in other models is to come up with an answer, then ask itself if the answer is right and how it could be wrong until the result is stable.

The seahorse emoji question is one that trips up a lot of models (it’s a Mandela effect thing where it doesn’t exist but lots of people remember it and as a consequence are firm that it’s real), I asked GLM 4.7 about it with deep think on and it wrote about two dozen paragraphs trying to think of everywhere a seahorse emoji could be hiding, if it was in a previous or upcoming standard, if maybe there was another emoji that might be mistaken for a seahorse, etc, etc. It eventually decided that it didn’t exist, double checked that it wasn’t missing anything, and gave an answer.

It was startlingly like stream of consciousness of someone experiencing the Mandela effect trying desperately to find evidence they were right, except it eventually gave up and realized the truth.

EDIT: Spelling. Really need to proofread when I do this kind of thing on my phone.

Everyone around you has a phone with a camera. Businesses and the government have additional cameras looking all over. The phone camera being less obvious and handsfree seems like an arbitrary choice of where to draw the line

They couldn’t do that from one photo though, they’d need several examples all believed to be the same guy. A swirl like that preserves some of the information and you can reverse it, but the lost data is lost. Do that for several photos and you can get enough preserved bits to piece something together.

Same idea for some other kinds of blurs or mosaics. Black boxes, not so much - you e got no data to work with, so anything you tried to reconstruct would be more or less entirely fantasy.

Why would you need oversight for access? All footage of all officers interactions with the public available to all citizens within say a few days of being filmed. That’s how you achieve oversight.

In fact, with storage capabilities nowadays, bodycams should ALWAYS be recording, period. Gotta go to the bathroom? Too damn bad. You’re a public servant. Trust the auditors to redact that if it comes to a court subpoena. You signed up for it. Extraordinary powers come with extraordinary sacrifices.

You don’t need to record them shitting, just make it explicit policy that any time police disable their body cams the default position for what happened is whatever is most favorable to members of the public and least favorable to officers. Full stop.

To be fair, I find the idea of a government outsourcing IT needs to entities under the sovereignty of foreign governments kind of fundamentally problematic to begin with.

So AI is a nice new technological tool in a big toolbox, not a technological and business revolution justifying the stock market valuations around it, investment money sunk into it or the huge amount of resources (such as electricity) used by it.

Specifically for Microsoft, there doesn’t really seem to be any area were MS’ core business value for customers gains from adding AI, in which case this “AI everywhere” strategy in Microsoft is an incredibly shit business choice that just burns money and damages brand value.

It’s a shiny new tool that is really powerful and flexible and everyone is trying to cram everywhere. Eventually, most of those attempts will collapse in failure, probably causing a recession and afterward the useful use cases will become part of how we all do things. AI is now where the internet was in the late 80s - just beyond the point where it’s not just some academics fiddling with it in research labs, but not in any way a mature technology.

Most gaming PCs from the 2020s can run a model locally though it might need to be a pruned one, so maybe a little farther along.

to prove the fucking obvious(that Tylenol doesn’t cause autism)

Eh, there were a few studies that showed a link including one of the largest ones. Now I get to move this from the “very unlikely but possible” column to the “almost certainly false” column.

There’s a pretty wide spectrum with autism, ranging from “a few quirks” to “nonverbal and more or less totally incapable of taking care of themselves, ever”.

Wouldn’t make a difference. We’ve got him for a bare minimum of at least another year and six days. Otherwise it would count as a term for Vance. They’ll Weekend at Bernie’s his bloated corpse for a year if they have to for that.

Given the Lemmy view on AI, I wonder how many folks are now uninstalling the game and demanding a refund because it’s suddenly transformed into “AI slop”? Or demanding it be delisted from Steam since they didn’t disclose their use of AI on Steam?

Meh, no process is perfect and sensitivity and specificity are often enemies. Basically, in a lot of cases the more sensitive you make a test to detect something, the more likely it is to accidentally catch false positives.

Sounds like they’ve vastly improved it’s ability to detect, hopefully that didn’t come with false detections for people running unusual hardware or software combinations.

They don’t give a fuck if murderers and armed robbers get away with their shit.

They care if murderers and robbers get away with their shit, they don’t care if murderers and robbers get away with your shit. Important distinction.