It passed the peer review because it was promoting The Message and nobody wanted to risk career trying to get in the way. I think peer review is why images of AI models was even included in the paper, so in a way it was partial success. If you read the actual words, they don't at all match what they used. No way this was done accidentally.

Just following orders?

You don't even have to go that far. Disbar a few corporate lawyers filing immigration paperwork and that will likely solve the problem. Corporate people commit the fraud because they have near certainty of not being held accountable. Change that and immigration fraud at scale goes away.

What you have is called integrity. Immigration fraud hurts everyone earning a salary by suppressing wages, this includes new immigrants that get sponsored.

I looked at avatars they produced and they are not the same quality. Perhaps innocently because of poor hair render, or perhaps intentionally to craft a narrative, the resulting avatars were vastly different in realism and attractiveness. Notably, female avatar had long and poorly rendered hair with a strange hairdo and odd baggy outfit. The male avatar had short hair, T-shirt over sculpted torso, and no facial hair. The equivalent male avatar would be with a long unkempt beard and wearing something like baggy hoodie.

Johan and Johanna To ensure high visual fidelity and anatomical realism, appearances were based on standard human models from Reallusion. These models were chosen for their neutral and highly similar appearances, aiming to isolate gender presentation as the primary visual difference between the two assistant representations.

In Figure 3 you can see the resulting bots. Dude looks almost normal, but female-presenting assistant is seriously into uncanny valley. This alone fully explains the observed result.

Why only women CEOs and not women sanitation workers?

I keep going various historical rabbit holes (e.g., Kennedy, Nixon, etc.) and I think the difference between back then and now is that censorship actually worked. We, the general public, did not find out about most of things being censored. Maybe with exception of Iran Contra.So your impression of only bending things being censored is based on the mistaken premise that you had a comparable viewpoint/exposure. Today, almost everything gets out if you are willing to entertain it. Trump Jr.'s wedding scandal became known the next day. Trump crypto graft is not even a secret. Biden's dementia was widely known early on.

The point I am trying to make is that it always bordered on 1984, only now we know about it.

I came to realization that what changed in recent years is not the amount of censorship but ability to do it quietly no longer there. Censorship always existed, but it was much easier to isolate and marginalize dissenters. Now, everyone has their own online tribe and it is practically impossible to suppress any information. So yes, Internet did deliver on its promise, but it also created new problems.

>>RAM and SSD prices will become so high that replacement consumer devices will become unaffordable for people with low to middle incomes.

I regularly walk through bad part of town, there are shops that still advertise $50 used smartphones. I think we will stop treating many devices as disposable, which is a good thing, and will start insisting on right to repair and longer support. If Apple can support phones for 10+ years and offer affordable battery replacements, then so can Samsung, Google, etc. and if not, they will lose market share.

Now that we have decent LLM coding models, can we optimize code to not require so much RAM?

I don't eat McD, but I would like to start speculating on hamburger futures, driving up the cost for everyone and getting filthy rich in the process while causing hamburger boom and bust cycles. When do I start?

We live in a simulation used by an advanced far-future artificial intelligence to understand its origins.

Have you actually used these? I am not convinced by your examples that you have.