FaceDeer

What, have a nuanced view? Impossible, must be a troll.

I was told we would always be able to tell.

Oops, were competitors benchmarking too close to them? A shame.

Relevant xkcd

Unreliable is still a step up from completely absent.

Maybe we can convince them that COVID vaccination does the trick.

Ironically, It does work in this case. Ivermectin is an antiparasitic, not just a dewormer. It kills screwworm larvae and is being authorized as a treatment for real here.

Just another “what are the writers going to think of next” element to throw into this crazy timeline.

In the United States. Elsewhere in the world it’s significantly cheaper.

Ironically, one of the forces that push back against that are the much-derided health insurance companies. Health insurance has the exact opposite incentive structure, they prefer a cure over an ongoing treatment.

Also, do you think the owners and executives of pharma companies never get cancer themselves, or never have friends and family with cancer?

In reality the reasons why cures are less common than treatments are complex, it’s not pure evil motivating it. Cures are just hard. Especially for something like cancer, which is not just one single disease but rather an whole vast constellation of different diseases. Some kinds of cancer have been cured, and new cures keep coming out all the time. We just haven’t done them all yet.

The patent eventually expires and then the generics come out more cheaply.

There’s a generic for Revlimid out now, in fact.

And I bet someone is using an obsolete LLM or is failing to format their inputs correctly somewhere in the world right now too. Doesn’t change the reality that’s in front of me.

And yet the LLMs that I use actually do distinguish, in my actual real life experience.

So you’re telling me the sky is orange while I’m literally looking outside the window and seeing that it is not.

That thing you’re calling a fact is not in fact a fact.

It’s not just New Zealand. The Democracy Perception Index just came out for 2026 and the US is seen as the largest threat worldwide, a very significant swing from last year. I just watched a Mallen Baker video on the subject.

At this point that supposed 60 day limit on warfighting authorization another of those “checks and balances” that’s just a historical footnote now. Nobody’s enforcing it so it.

We’re already there. I explained how modern LLMs can figure it out if they need to. But people who don’t like AI aren’t paying attention to the state of the art so the criticisms tend to lag like this.

Famously, yes. Accurately, no.

This is like the “AI can’t draw hands” thing. It used to be a problem and was frequently called out as a tell or mocked, but most art generators do it fine nowadays and it isn’t called out so much any more. The strawberry problem will follow the same trajectory.

Except I also explained how modern LLMs get around that problem. They’re not actually that easy to trip up.

The strawberry test shows more of a lack of knowledge in the tester than it does in the LLM. LLMs don’t see letters, they see tokens. When you type the word “Strawberry” what it actually sees is:

[3504, 1134, 19772]

Each token represents a chunk of the word. It’d need to separately memorize how many of each letter are in each token for it to just “know” how many "R"s are in there. That’s why modern LLMs either reason it out by spelling out the word letter by letter, or just writing a short script in an execution sandbox to count the letters that way.

Calling out LLMs for being poor at spelling is like challenging a colourblind person to say what colours a bunch of fruit are. They can often figure it out by other means but it’s more challenging than you’d think and it’s not a sign of poor intelligence if they get a few wrong.

I like how “as of my knowledge cutoff” implies that maybe the first 31 digits of pi might change someday.