Mimir Systems

This paper shows that today’s AI detectors are not trustworthy enough to decide whether a student used AI. The authors built 3 large datasets from real pre-GenAI student work, paired them with AI versions, and ran 13 commercial and open tools on more than 280,000 samples. They found that detectors did somewhat better on long theses, but they broke down on short coursework and especially on engineering code, and STEM writing was more likely to be flagged unfairly because technical writing often sounds formulaic. --- sciencedirect. com/science/article/abs/pii/S0360131526000540

You know how sometimes you ask around the lab and find the one person who knows the answer to a specific question, and how the rest of the time you spend days finding three papers without even getting complete answers? Mimir is that for you, except for everything, all the time - 1M+ papers - Responds in seconds - Always available 24/7

You know how sometimes you ask around the lab and find the one person who knows the answer to a specific question, and how the rest of the time you spend days finding three papers without even getting complete answers? Mimir is that for you, except for everything, all the time - 1M+ papers - Responds in seconds - Always available 24/7