ah I get what you’re saying., thanks! “Good” means that what the machine outputs should be statistically similar (based on comparing billions of parameters) to the provided training data, so if the training data gradually gains more examples of e.g. noses being attached to the wrong side of the head, the model also grows more likely to generate similar output.

I only have a limited and basic understanding of Machine Learning, but doesn’t training models basically work like: “you, machine, spit out several versions of stuff and I, programmer, give you a way of evaluating how ‘good’ they are, so over time you ‘learn’ to generate better stuff”? Theoretically giving a newer model the output of a previous one should improve on the result, if the new model has a way of evaluating “improved”.

If I feed a ML model with pictures of eldritch beings and tell them that “this is what a human face looks like” I don’t think it’s surprising that quality deteriorates. What am I missing?

I can’t say if this specific one is authentic but I would not be surprised, as my company does the same: branded pride flags, t-shirts for their LGBTQ+ employees. And I’m torn on this, because I feel a bit bad about the branding aspect of it, but on the other hand the company putting their name to openly support all sexual and gender identities does seem like a good thing to do…

Yes, the irony of a company that produces more efficient ways of killing people being concerned with social issues does not escape me. But there are employees of the company who apparently support both (and good for them)

I was as surprised as you to see Apple mentioned as supporting the bill, in the article. Thanks for pointing me to Fairphone, seems amazing!

So far I’ve mostly seen iPhone 15 panned for lack of innovative features, but if it turns out that it’s actually easier to repair (as Apple is saying) it would be a killer feature for me

I would be surprised if Whatsapp tried to implement its own version of Telegram’s, Signal’s and every other messaging app’s protocol to “talk” to all of these other apps. I bet they will provide an API to interoperate with Whatsapp that these other clients may (or may not) choose to implement, in order to send their messages to Whatsapp users.

In that scenario it would up to Signal (if they implement this) to choose how to display to their users that they are sending a message to someone who’s using Whatsapp, or to create options for users who want to disable this completely.

I agree with you, and that’s what I choose to think when I feel like the “best” version of me.

But there are moments (or a part of me) that has a way more violent disposition and feels differently about people who do terrible thing.

I’m a very calm person and not at all violent so please don’t report me to the police on the base of these posts… That violent part of me is small and weak, but I just think it’s important to acknowledge it because it’s also the part that makes me recognize that a rapist or a murderer is a person like me and that it might be me, with the wrong set of circumstances, life choices and frame of mind.

He mentions rapists and murderers. I choose to think that they don’t deserve to be given a gruesome death (or a death sentence at all for that matter) but I’d be lying if I said I have never felt that way when reading about some f-ed up stuff on the news. Maybe I’m also higher on the Hitler scale than I thought I was.

Shit, I’m sorry, it was insensitive of me to make that joke (well, half-joke)

Well, they probably didn’t do it very scientifically but if they could think of it and the tools existed, someone in history is likely to have tried it as a method for killing people.

Impaled people, for instance, could allegedly take days to die. Being slowly eaten by ants or rats sounds pretty painful too

There’s one called “life” that is pretty cruel too. It might take anywhere from seconds to more than a hundred years for it to eventually kill you and some.people get to experience a lot of pain throughout the experience.

Well after seeing Teams that would actually be a decent line of defense… “we aren’t competent enough to build something that works as intended”

No, you see, we didn’t intend to ask…

Zorr’ak notes, page 2:

Earthlings’ nutrition consists mainly of elongated pieces of synthetic rubber. These can’t be fully digested, so consuming the nutrients they contain is achieved by inserting them repeatedly in one or more of the mouths that Earthlings have in different parts of their anatomy.

Zorr’ak notes, page 3:

Several attempts to feed the Earthling specimen we collected have failed. We are releasing it to avoid it dying from starvation

NY Post, page 3 headline:

Texas man claims he was abducted by aliens and anal-probed

I’m assuming by “need my rating” you mean “need to be rated positively” (and not “need my honest feedback so they can improve their product”).

If so, I do that too, but I think the article has a point that a 5* review can now be more like a vote of “I wish more people bought this/supported this company” than “this product is really top notch”. This is much more useful to companies than it is to other buyers.

I can see many many examples of how bad Microsoft and Google can be. However this one I honestly don’t understand: how’s Google supporting Mozilla’s competing product anti- competitive? Are they forcing Mozilla to do things they don’t want in return?

I am a Firefox uaer and on every install on a new machine (or phone) I switch the default search engine to duckduckgo. But for good or for bad Google is the search engine most people use (and would use on FF too even if it wasn’t the default). I don’t think Google needs to force Firefox 3%-ish market share to use their search engine.

Sorry if I gave you the impression that I was trying to disagree with you. I just piggy-backed on your comment and sort of continued it. If you read them one after the other as one comment (at least iny head), they seem to flow well

I’m not even remotely a legal expert and I don’t know what type of popup that is but I think the anti-competitive piece is “could Google use the same technique to push the user to switch to google search on Edge or not?”.

If this was an ad from a web page OP had opened or from the game and if clicking “Yes” only directed the user to a site with instructions on how to switch default search engine on Chrome, then yes, obnoxious but probably fair. Google could strike a deal with the game developers to push their search engine to Edge users or buy an ad. Someone writing a new browser or search engine will probably have considerably less money than Google but could reasonably do something similar to try and gain market share.

On the other hand, if that popup comes from Windows itself and especially if clicking “Yes” directly changes Chrome’s settings, then this is Microsoft using their ubiquitous (on desktops) OS to nudge more users to switch a competitor’s browser to their own search engine. Google, or even less a new competitor. would probably not have the same type of OS-level access to switch the settings of a different browser.

it doesn’t even look at the smaller picture. LLMs build sentences by looking at what’s most statistically likely to follow the part of the sentence they have already built (based on the most frequent combinations from their training data). If they start with “Hitler was effective” LLMs don’t make any ethical consideration at all… they just look at how to end that sentence in the most statistically convincing imitation of human language that they can.

Guardrails are built by painstakingly trying to add ad-hoc rules not to generate “combinations that contain these words” or “sequences of words like these”. They are easily bypassed by asking for the same concept in another way that wasn’t explicitly disabled, because there’s no “concept” to LLMs, just combination of words.

For the US in the list of countries starting with M, maybe too many 'Murica memes in the training set?

If you didn’t like it, you shouldn’t have let your laptop walk around dressed so provocatively