1. Deep learning is being used in medicine for tasks like classification, segmentation, and detection using convolutional neural networks. Google has developed algorithms for diabetic retinopathy detection and cancer metastasis detection with high accuracy. 2. Unsupervised learning techniques like generative adversarial networks show promise for generating medical images but have challenges around validation. 3. Concerns with deep learning in medicine include the need for large labeled datasets, validating models across different patient populations and settings, legal and responsibility issues, and discrepancies between clinical and general populations. 4. Future areas of focus include generative adversarial networks and reinforcement learning. Cooperation between researchers and doctors will be important to address challenges around credibility and validation of deep learning models