The concept table in your mind does not specify its length or width, mass or its number of legs. In fact even if you look at a post-modernist design of a table you would know what to do with it: i.e. place things on it. Randroids like me call this concept measurement omission. It is the process by which the brain forms concepts from precepts/other concepts. Many experimentally determined models that describe the functioning of the brain, describes this process in detail. To be less vague, measurement omission is a form of deep lossy compression, which compresses perceptual data multiple times, successfully omitting some aspect of measurement from each level of compression. The current state of artificial intelligence / machine learning research has not achieved the level of measurement omission like what our brains are capable of. Most AI techniques identify thresholds of some sort in-order to identify perceptual artifacts as members of a particular set. I believe the next step in AI research is to achieve the holy grail of this ultimate measurement omission which does not even rely on thresholds defined on percepts. Certain possible anomalous concept formations must be avoided to form concepts which correspond with reality realistically i.e. in a manner which is conducive to sustaining to concept formation process. For example, there are documented cases of patients with psychosis, who have attempted to form the concept "encirclists" which correspond to pictures of Jesus with a halo and cigarettes with rings on it. i.e. their brain erroneously tried to omit the presence of something circular from these concepts/objects to form another (meta)concept called "encirclists". These concepts are erroneous because they result from faulty induction. The correct reality based induction is described in the book "The Logical Leap: Induction in Physics". Too lazy to describe it here.
So what if we actually achieve an artificial inductive process? Will it be like minds we have? Not quite. Because the human mind is not just induction engine which omits measurements to form concepts from precepts. Randroids believe it possesses volition.
While Randroids only have vague descriptions, I personally believe it has something to with partial solutions to Turing's halting problem re-stated for finite sets of deterministic process. But this margin is too narrow to contain that remarkable insight.