3 ms·Inducing anxiety in large language models increases exploration and bias4 points by Borrible 3y ago