 Okay, we are at the end of the semester almost, well apart from your projects. How can I summarize this in sort of one slide of what we've seen today? So far this entire semester has focused on what I would call narrow tasks. Recognizing specific objects, doing machine translation or speech recognition or game playing or drug design. We haven't covered that but it's cool. These are all specific tasks that deep learning systems do really well and we've seen that by using pre-training, by using auto encoders, by using transformers, by building it in variances like filters for CNNs, we can train these things on much smaller datasets than were required before and do really well. So superhuman performance on a number of very narrow focused tasks, playing go, recognizing images. Awesome! What's left? Well lots more of those for all the big and small companies of the world, but also a whole different area of artificial general intelligence. Can we get computers, deep learning systems that can generalize that when they play one game they can learn to play a different one. When they learn one language they're faster than their language, that they can understand what's happening in the stories and the world and predict what would have happened if John had walked out from the restaurant without paying. Right? Counterfactual questions. And I think those will probably need some form of modularity. Will it be capsules? Will it be RIMs? I don't know but something like that, something that's probably neuro-symbolic, some way of breaking the world into modules that have sparse interactions, something that's going to capture some form of causality? We'll see, hopefully you will see. I've provided one fun link from Josh Tenenbaum, one of my favorite cognitive scientists, who has a research program trying to push these sort of things at least in small toy worlds like the cleverer system that we saw. So thank you. It's been a challenging semester but I really appreciate your sticking with us. You guys have been a great class. I am so looking forward to seeing your final projects and do stay in touch with me and with Conrad as you go off the world and use all these techniques. I love it when students come back and say, hey I'm having this cool problem at my company. How might I do that? So do stay in touch. Thank you and I'm sure I will see you all talking about the projects. Have a great rest of your semester.