To not invade the brain by using external devices to provide brain-to-computer interfacing is obviously desirable. But will it be efficacious? Invasive techniques provide higher resolution and hence provide more precise control, but non-invasive devices that utilize machine learning may provide control that approaches the resolution and precision of invasive devices. The answer at this point in time appears to be that, for smooth movement and conversational speech, invasive devices are needed. For all other needs, non-invasive techniques should suffice.