To progress in the search for human-level artificial intelligence, we need to focus on areas of research that truly matter. This means avoiding our inherent biases as researchers, as well as biases imposed upon us by our environment.
History has taught us repeatedly that fashionable and trendy approaches, albeit exciting, are frequently short-sighted and possibly detrimental to the long-term progress in the field. For this reason, as a community we should strive to distance ourselves from our biases and open up to the 'bigger picture', especially from a long-term perspective.
This should allow us to compare and contrast various approaches to solving human-level artificial intelligence and identify the aforementioned potential dead ends and ideally highlight open research problems that we should focus on now, rather than be swayed by current trends in the field.
The institute continually maps the field and identifies such areas and publishes them. The wider community is then able to openly discuss these in our forum and filter out unnecessary distractions and branches of research that are less likely to lead us to the ultimate goal.