Hacker News Reader

Comment by svnt | Hacker News Reader

The more common english term is “held back” instead of gated.

And nvidia probably doesn’t want to say “probabilistic responses including hallucinations preclude reliable ai analytics”

As it is the sentence communicates very little. Things are unreliable for a lot of different reasons. Hardware? Software? Users? Intentional and required use of RNGs? Claude presumes the reader has all the same weights activated.

View on HN

I've used the word "Gate" in this context.

People are really seeing 'AI' in everything now. You can't tell.

Humans are now hallucinating what is, or is not, AI.

Using gate as a verb is extremely rare in American English, though perhaps more common in British English. I would hazard a guess that very few people who aren't exposed to Claude on a regular basis would think that was anything other than gibberish.

So you're claiming a Claude-shaped readability gate? /s

LOL

People are "gatekeeping" how we use the word "gate".

Using gate as a verb is perfectly fine and I see it all the time where I grew up in America. The problem with the sentence is deeper than word choice. It is like politician language in that it technically says what Claude is attempting to say without ever nailing down the precise context, so that it can be interpreted a million different ways and all would be correct, if you read it right. Which means it says nothing.

I am convinced this is a result of RLVR and synthetic data. LLMs optimize not for human readers, but for passing the LLM evaluators in their training runs, and end up developing sophistry.

I've been screening some twitch streamers and noticed with some alarm that an unexpected number of seemingly healthy American streamers can't read, or very barely can.

The game I was using for this uses a standard memo pads mechanic to fill in the story and many were barely stumbling through it in their own native language.

So it's my impression that this is not entirely Claude's fault.

Conservatively, I'll blame it on "dyslexia".

The line follows: "...a revenue metric answered faster is worth nothing if the number is wrong.". Are you seriously suggesting a human wrote this?

Its tricky with corporate speak. This was released by a big corp. Marketing slop existed before AI. I can't tell. But maybe that is the point, humans have written some convoluted sentences so can't just assume by how it sounds.

Seems like an extremely simple sentence to me, and is an example explaining what previous commenters seem to have had trouble understanding.

Reply on news.ycombinator.com