Tokenize the same sentence by letters instead of words, and its perplexity falls 66 percent
Perplexity scores a model's surprise at real text, but the same sentence scores differently depending only on the tokenizer, the blind spot that let a 2025 study catch Llama 3 favoring ungrammatical Japanese over grammatical Japanese.
The slogan ‘stochastic parrots’ outgrew the argument that coined it
The 2021 paper grounds two of its four risks in studies it cites, and leaves the other two, an argument about research priorities and the philosophy behind its slogan, without a citation under either.
Intelligence says nothing about what a mind wants
Bostrom's 2012 argument treats intelligence and goals as separate variables, so a mind can be as capable as you like while pursuing nothing but paperclips.
Steven Schwartz's ChatGPT brief cited six court cases that didn't exist
Judge P. Kevin Castel fined Steven Schwartz, Peter LoDuca, and their firm $5,000 in June 2023 for filing the fabrications and then telling the court, under oath, that they checked out.
A model given the right document can still score worse than one given none at all
In a 2023 study, GPT-3.5-Turbo answered correctly 53.8% of the time when the right document sat in the middle of 20 documents, worse than the 56.1% it scored with no documents at all.