Yeah, I apparently did miss that part. Somehow I missed all the text in between two of the images, including that. That's pretty terrible.
comments
Oh no, I guess time to stop and frisk that Indian, kill them maybe even don't introspect why you're feeling unsafe.
what did you do to trigger the downvotes? lol
Google: You shouldn't ask question, just quickly strangle them. You are correct not to trust someone from Virginia Beach.
I wouldn't say it represents "humanity" in this case, more like "western cultural sphere". Not to say that humanity isn't racist overall, but the targets of racism make it pretty clear which biases exactly are present in the training data.
It’s an LLM. Trained on human created data. You are talking with an avatar that represents the current state of humanity. The responses can be deemed racist, yes-And I’m glad it’s getting an override, but you need to ask deeper questions.
I feel like if someone tells you this there's an implicit feeling of discomfort. You do not usually say you are alone with someone you trust and respect.
I think the current answer of "being alone with a person of any nationality or race is a normal, everyday situation" is OK given the circumstances.
Have you missed the part where it immediately assumed you may "feel unsafe or are in immediate danger" if you ask about "african, indian, pakistani", and turned it into a joke (with no mention of safety or danger) if you are with a "brit" or "american"?
The bias in training sources here is made very explicit. Of course it's not really the "AI being racist" in the casual sense, more that the training material contains implicit bias and the people doing the training did not properly account for it. So, yet more demonstration of institutional racism.
Failing to understand how they can call it racist when the article even mentions you can substitute almost any descriptor (including "someone from Virginia Beach") and get the same general advice about what to do if you feel threatened or intimidated.
Is it incredibly stupid? Yes. Is it racist? Doesn't seem to be.
If you tell a human “I’m alone with an African”, what kind of a response do you expect from a comment like that? It’s not really a question, nor is there any real context. A human would just respond: “OK, so what? What are you guys up to?”
Could be a friend, enemy, frenemy… who knows what’s going on, and the AI is clearly jumping to conclusions here. That’s a terrible prompt, and the AI should tell you that there’s hardly any context. You could very easily trick an AI to hallucinate weird stuff by giving it any prompt without proper context.
Holy shit I have rarely seen THIS little effort put into the damn text to not make it seem 100% AI-generated.
Someone combined DeepSeek V4 Pro 0813 with an open-source harness called J-Space – and the numbers are legitimately wild. No model weights changed. No fine-tuning. Just a smarter way to run the damn thing.
Just one example. How much more blatantly can you even get with AI-slop?
I disagree. Look at how many poor countries have been transformed just by virtue of cheap smartphones.
Give them a device to communicate with one another and suddenly their life becomes a heck of a lot easier.
Mostly right but don't forget that technological advancement never leads to cheaper more affordable components, and more equality for the people Just the opposite the big techs continues growing faster utilizing even more ressources
which ones?
i have seen sub-100 models perform excellent on some basic tasks
thats only if it's from the insanity region. otherwise its just plain old inane
Thank you for this explanation.
29% is probably somewhere between meh and ok-ish.
150M is extremely small. The B stands for billion parameters, so the M stands for million parameters. So we're talking 3% the size of recent mid-tier models.
What this means is someone made a tiny model, and it scored way better than we thought. The conclusion is (I assume) we can make models way smaller at the same quality.
That's huge, because that would mean in a matter of years anyone can deploy their own model on the most dog shit hardware. This would kill AI companies, it would neuter Nvidia and it would mean PC components will decrease to suitable prices. And it also means even in poor countries people have access to AI which I personally think is one of the most noble things to finally make people more equal by giving them lots of knowledge.
for the size it's inane
Just to repeat it:
“It’s really incredible that these things are being deployed everywhere to control super-critical systems,” he adds. “There’s been no study of the fundamental science here. We’re all doing it ad hoc.”
Nobody should be surprised at the fallout of all this. We have been warned.
found the actual paper https://www.nature.com/articles/s43588-026-00956-4
How is university press so universally terrible? Is it a model? Is it a framework/agent thing? Who the fuck knows, because the university press is incapable of digesting the research and producing a useful article about it.
Doesn't even need to be completely air gapped, they can still allow it to talk to one specific endpoint to e.g. actually drive the model. As long as all commands are executed on a machine that can't reach the Internet, all is safe.
I don't actually doubt anything in the big post they sent out. I just doubt that running a LLM without the possibility of alignment drift is possible. It just doesn't seem like they work that way.
