When scholars explore new ideas, it seems instinctive to share their thinking. Last year computer science professors Joseph Halpern and Bart Selman became co-principal investigators for a nationwide project to ensure that robots and artificial intelligences (AIs) will act in ways beneficial to humans. "We thought we should get undergrads involved in thinking about these ideas," Selman said. Thus was born a new course, CS 4732, Ethical and Social Issues in AI, about how robots and artificial intelligences may change our world, and what we ought to be doing about it. "These undergrads may be directly involved in developing the software behind these systems," Selman added, noting that many may go on to jobs with Google, Tesla and other major players.
Automation is a complex topic that has received a lot more focus in recent years. Many experts have been predicting that many unskilled jobs could soon be replaced by automatons. However, the impact could be much further reaching than that. Newer reports suggest that AI algorithms could soon start replicating computer code, which will possibly put coders out of work. These predictions seem very far-fetched to some people.
" Abstract Recently, we have seen some promising developments in using machine learning techniques to speed up search methods. For example, Boyan and Moore introduced the STAGE algorithm to learn how to reach good starting states for local search methods. Also, Dietterich et al. have proposed a range of reinforcement learning methods to discover heuristics for solving job shop scheduling problems. Despite these promising developments, these techniques have not been incorporated in the latest combinatorial solvers, such as SAT and CSP engines. On the other hand, techniques th a machine learning flavor but not based on traditional machine learning techniques, such as clause learning and clause weighing, have been shown to be of direct practical use.