Human-AI Interaction in Symbolic Problem Solving

Despite the increasing need for computer assistance in solving problems involving complex systems and large amount of data, professional mathematicians, scientists, and engineers currently avoid the use of computer algebra systems during creative problem-solving phases of their work due to problems with transparency, familiarity, and inflexibility in input. I have designed and prototyped a new approach to interaction with computer algebra systems that is compatible with current working styles, flexible in its input and output. I propose a user study to validate this tool, and tool extensions to allow creative problem solvers to interactively define their own notation as they work.