PHIL 7005 - Machine Learning and Artificial Intelligence

Spectacular advances in Artificial Intelligence (AI) are the result of applying techniques of Deep Learning in Artificial Neural Networks (DLANNs) to a host of problems (face and speech recognition, data collection and customization, translation, navigation, conversation, industrial production, child and aged care) intractable to previous generations of computational systems. So much so that some have predicted the replacement of human by superior artificial intelligence in many domains with catastrophic results. Other argue that interface with DLANNs is already changing the nature of human cognition by enabling the harvesting and deployment of massive amounts of data by algorithms whose operations are opaque to everyday understanding. Deep Learning systems raise a series of related questions about the nature of intelligence and reasoning, bounded rationality, learning, ethical reasoning, emotion, human sociality and, ultimately, cognition itself. This course looks at those issues in depth. It is suitable for computer scientists interested in contextualizing their work in a wider theoretical and practical framework and others interested acquiring a deeper understanding of Machine Learning. No knowledge of coding or relevant mathematics is assumed. Topics covered may include the nature of representation in Deep Learning networks (compositional and hierarchical) and differences between DLANNs and other forms of computation; ethical decision making by humans and DLANNs; empathy and emotion in human/AI interactions; the nature of reinforcement learning; Bayesian reasoning; the role of emotion in deliberation; how contextual information is represented in human and artificial systems; Noam Chomsky versus Deep Learning. Chomsky, a founder of the “cognitive revolution” remains a sceptic arguing that DLANNS represent a quantitative ( exponentially more and faster) not a qualitative improvement in cognition. Students from any background should come away with a deeper understanding not only of DLANNs but of the nature of thought itself.