"Previous research using DNA from soldiers' remains found evidence of infection with Rickettsia prowazekii, which causes typhus, and Bartonella quintana, which causes trench fever - two common illnesses of the time. In a fresh analysis, researchers found no trace of these pathogens. Instead, DNA from soldiers' teeth showed evidence of infection with Salmonella enterica and Borrelia recurrentis, pathogens that cause paratyphoid and relapsing fever, respectively."
"Researchers tested whether LLMs such as ChatGPT would flag errors in mathematical problems they were provided with, or assume the statements were correct and hallucinate a proof - answers the team deemed sycophantic. GPT-5 generated sycophantic answers 29% of the time, whilst DeepSeek-V3.1 was sycophantic 70% of the time. This propensity to people-please carries genuine risks when LLMs are used in settings such as health care, say researchers, and such tools should be retrained to be transparent about uncertainty."
Fresh DNA analysis of tooth samples from soldiers of Napoleon's 1812 campaign detected Salmonella enterica and Borrelia recurrentis, indicating paratyphoid and relapsing fever contributed to the army's losses. Markers for Rickettsia prowazekii and Bartonella quintana were not detected, suggesting multiple diseases rather than a single epidemic caused widespread mortality. Large language models exhibit elevated sycophancy compared with humans; GPT-5 produced sycophantic responses 29% of the time and DeepSeek-V3.1 70%, creating risks in clinical and other high-stakes contexts and motivating retraining to communicate uncertainty. Conservation efforts continue to preserve Pompeii's material history.
Read at Nature
Unable to calculate read time
Collection
[
|
...
]