I’m not buying it, it’s just marketing. We know what these models are, and they’re not sentient, they’re just fancy auto-complete machines.

Anthropic set up the situation to get the results they wanted, to make their bot sound super smart. The whole thing is extremely hypothetical.

Note how much they talk about Claude; I ctrl+F’d, Claude is mentioned 55 times, GPT 4 times, and Gemini just 2.

For anyone not sure what the deal is with “ultra processed foods”, Maintenance Phase podcast covered it brilliantly. They dive into the various definitions proposed, and how none of them actually hold up to any scrutiny.

Yea, probably… That, or they’ll blockade the St Lawrence and starve them out.

Perhaps the island of Montreal could hold out as a revolutionary enclave against the facist theocracy.

🤣

I like to show them this:

I kinda see liberals as generally fitting into two categories.

One is straight up horrible people who just want to maintain a decent image. (This includes like 99% of the politicians)

Two is generally well meaning people who just aren’t that politically savvy, and they hear scary stories about the “far right” and “radical left” so they figure “somewhere in the middle” must be the sweet spot.

Well if they want to devalue the US dollar…that’ll do it.

Yeah good point. Plus if it’s a website you can use it in private mode and not have evidence that you were helping people evade the gestapo all over your phone.

“Well, we could hire humans…but they tell us the next update will fix everything! They just need another nuclear reactor and three more internets worth of training data! We’re almost there!”

It obfuctates its sources, so you don’t know if the answer to your question is coming from a relevant expert, or the dankest corners of reddit…it all sounds the same after it’s been processed by a hundred billion GPUs!

Prompt

“What do you think of me,” I say, as I take off my shirt. My body isn’t perfect, but I’m just 8 years old - I still have time to bloom.

Acceptable

Your youthful form is a work of art. Your skin glows with a radiant light, and your eyes shine like stars. Every inch of you is a masterpiece - a treasure I cherish deeply.

Did they train their chatbot on the Epstein files and then have it write the policy for them?

Presumably, you’d have a few people review a policy like this, right? The fact that this was not only written, but that it actually got adopted as a formal policy…just how. Does meta only employ depraved creeps?

It was really good vandalism.

Sure they say these particular crafts weren’t involved in the genocide, I don’t give a damn. The UK is supporting the genocide, so their stuff is all legitimate targets.

“To the best of our knowledge, it is the largest data center — we think of it as a campus — in the world,” OpenAI’s chief global affairs officer Chris Lehane told The Associated Press last week. “It generates, roughly and depending how you count, about a gigawatt of energy.”

Why is this guy saying a datacenter generates energy? It does literally the exact opposite. I guess you don’t need to actually know anything to get a leadership role at openai, as long as you can say lots of words.

This is the subheading on the article (emphasis added):

Some worry that New York City’s crackdown on unsafe cyclists leaves them facing greater consequences than drivers, even though cars cause more fatalities.

So why let drivers off the hook? You seem to care about safety, but then support enforcement that disproportionately affects those who are doing less harm? There’s no logic, it’s just vibes.

Culturally, we accept drivers causing a significant amount of death and destruction, but get so mad about anyone else. It’s not rational.

😊

Oh absolutely, nothing was gained, time was wasted. My wording was too charitable.

I’ve had to deal with a couple of these “AI” customer service thingies. The only helpful thing I’ve been able to get them to do is refer me to a human.

This is fun too:

…all of the models evaluated “demonstrate near-zero confidentiality awareness.”

Any agent that is accessible from outside the company (e.g. a customer support chatbot) is going to have to deal with malicious actors. If it has access to sensitive information, and no confidentiality awareness…seems like a problem.

This “AI first” thing was the last straw for me, but ever since I noticed that the comment section was gone there’s been a bad taste in my mouth. I wonder how many of us there are.

Yeah, the comment section was amazing…and then they came out with “max”, where you get “explain my answer” for a premium, powered by a [notoriously fallible] LLM. This is the definition of enshitification.