CrowdSense is an effort to better understand what types of questions people consider common sense. The AI model used here is RoBERTa (Liu et al., 2019) trained on CommonsenseQA and a set of questions collected using this app.
Each question you issue will be stored, and verified by 2 different verifiers. Only if the verifiers mark the question as common sense and agree on the answer, will the question be used for training the model.
The AI model is updated every 2 days on average. The result on the AI accuracy gauge is of the last model, evaluated on user questions it was not trained on. Let's see who's winning! (33% is random)
Please try to add creative commonsense questions that you think a child under the age of 7, anywhere around the world, would be able to answer. For inspiration please take a look at the recent examples section, that contains questions the AI model did not answer correctly.