A considerable portion of our research efforts currently targets fluid intelligence as a form of problem-solving ability independent of previously encountered examples.
A clean and expertless slate
Our preferred approach to fluid intelligence is to have as little human bias as possible.
This is a postulate driven by two key factors. First, the emergence of fluid intelligence is currently not understood very well. Second, it seems reasonable to assume that conventional computer systems are not well suited for fluid intelligence, simply because they were developed for other purposes.
In other words, we are concerned that the primary contribution a human expert can bring to the table in this area is bias that has detrimental effects. Such bias might completely derail the research or just lower the chances of finding the optimal trajectory of advance.
It is worth noting that in line with these considerations, we do not treat 'human level' with any special significance in relation to intelligence or as a trait exclusive to humans.
Our Fluid Intelligence Project is currently aiming at a machine-based development (simulation) of artificial fluid intelligence using conventional means of hardware and software. In this stage of our research, we are not particularly focused on the level of fluid intelligence, only its consistent presence. And given the regular hardware environment we are working with, we can also set aside the question of sentience.
To make this research successful and have a clean slate as described above, we have developed a 'bit' suitable for evolutionary development and having the least amount of pre-determined factors possible.
See our Fluid IQ Gym, designed to help visualise the evolution of automatic challenges used as inputs for some of our fluid IQ simulation research.
A longer-term research objective at FINAL Labs™ is to identify a medium better suited for hosting fluid intelligence.