A New Research from Google DeepMind Challenges the Effectiveness of Unsupervised Machine Learning Methods in Knowledge Elicitation from Large Language Models