AI-generated writing is frustrating not merely because it has recognizable habits, such as “delve” or frequent em dashes, but because it makes readers work harder to understand important ideas. Casual AI content can often be ignored; workplace messages and code-review descriptions cannot, especially when the underlying subject is already complicated. The argument is that language models remain poor writers, in part because clarity is harder to reward directly than capabilities such as coding, which also offer more immediate economic value. Clear writing is a responsibility to readers, not just a matter of taste: people broadly agree on what makes communication easier to follow.
A comparison of AI and human writing found unclear communication in 54% of AI-written articles versus 10% of human articles, and in 78% of Claude Code discussions versus 11% of Stack Overflow posts. The analysis groups the problems into three faults: hiding the people or things acting and the actions they perform; using vague or unnatural wording, including awkward combinations of terms; and making empty statements or burying the main point under misplaced emphasis. In coding discussions, vague wording was especially common. The results also distinguish clarity from AI detection: detectors identified nearly all AI samples, but that does not explain why they are difficult to read. The practical standard is whether readers can quickly identify what happened, what matters, and what to do next.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Researchers tested 16 LLMs on all 55 TIS-100 puzzles, giving each one attempt and time to refine its code. The models found 133 new performance tradeoffs across 18 puzzles and set three outright cycle records, though human players improved on most solutions within days. (tis-100.wincon.dev)
Australia is considering a new criminal offence for AI companies that fail to take reasonable steps to prevent harm caused by their models. The government is also weighing an AI safety watchdog with power to impose significant fines, following a hack of a Medicare statistics portal involving OpenAI. (abc.net.au)
AI is increasingly becoming a gateway through which people find information. Audiences still resist having AI produce their news. (niemanlab.org)
Nvidia holds a central position in the AI industry, making disruption to the company potentially catastrophic for the wider ecosystem. The race for advanced AI is also intensifying economic and geopolitical conflict, while concentrated corporate power and infrastructure face growing political resistance. (france24.com)
LLM inference can require reading about 26 GB of model weights to generate each token from a 13-billion-parameter model. The explanation shows how model size drives memory bandwidth demands during text generation. (youtube.com)
Deletion Tuning matches full fine-tuning across three LLM post-training methods by zeroing fewer than 0.05% of model weights. Its compact masks also preserve prior-task accuracy better than full fine-tuning, LoRA, and random sparse tuning. (arxiv.org)
Today's best Hacker News stories, summarized and screenshotted, one email a day.