The field of spoken language technologies (SLT) has been dramatically transformed in the last 10 years by the wide-spread proliferation of deep learning techniques, including systems for speaker identification and diarization, speech transcription and synthesis, speech translation, and parts of spoken dialog systems. Is this the end of the road for new methods in SLT? Is knowing how to train and deploy large neural models using a mixture of self-supervised pre-training and task-dependent supervised training all one needs to know in the future? Or is fundamental knowledge about speech and language production, perception, and cognition necessary for the next quantum leap in capabilities? Is there still a role for Bayes decision theory? What skills should new researchers to the field acquire? What should academia focus on and what is best left to industry? These and many such questions are being asked at all levels in our field, from the newest undergraduate entrants to the senior-most leaders. This panel discussion will be based on prior interviews with a number of senior researchers in the field. Panelists will debate some of their more thought provoking opinions, formulate and defend some of their own, and entertain a few reactions and responses from the audience.