In the introduction, I have presented the notion that concepts like "money", "investment theory", "wealth building", "financial success vs. failure", etc., can be understood within a context that is analogous to the design principles that are associated with nuclear weapons vs. nuclear power vs. stellar ignition, and so on.

I postulate, therefore, that a successful AI will need a high-quality world model, which, IMHO would have to include some type of framework for such things as relativity theory, as well as for logical constructs concerning some type of quantum loop gravity, etc.  Such a framework would give also the program the ability to understand such things as how to braid hair, "understand" what DNA is, "imagine" Mobius loops, and all other such things that might otherwise require fluency with some type of 3D-CAD software – at least insofar as it is necessary to provide the AI the ability to formulate connection maps which are diffeomorphic to the geometrization of space-time. 

Thus, is always interesting to contemplate the possibility that the Universe is actually just data, and that we are, of course, living in a simulation.  Since this log entry pertains to some kind of contest, on a system that calls itself Hackday.io, that must of course present some kind of conundrum - and a very important one at that - and that is - how to "hack the system", and with what tools and materials shall we make for our best efforts.  Do we fire up the De Lorean and head to the local hardware store and buy a few kilos of Plutonium and build a time machine?  Or should we build a new type of quantum computer that can hopefully crack the Satoshi hoard?  One of my favorite computers of all time, in science fiction, that is, was the machine "Orac" from the 70's British Sci-Fi series Blake's 7.  Orac had some nice features, such as the ability to use (I think) tachyonic sub-space energy fields to control other computers.  Maybe I will finally build Orac.  Or maybe not quite yet.  Rest assured that there WILL be some kind of hardware associated with this project, even if it turns out to be a simulation of a single quantum bit computer - it will at LEAST be able to sit there and blink!

The World Model still comes to mind.  I really want to build something, eventually, that employs some form of speech recognition, and which can at least attempt to engage in "Eliza" type conversations with its users.  To some extent, therefore, I suspect that the "power of narrative" must eventually enter into the equation, since a carefully written narrative can provide a VERY convincing interactive experience for the user, that is if one wants to believe that they are "interacting with some kind of sentient AI".  Likewise, if I could introduce some kind of simulation of at least one aspect of any type of quantum computer - then that would also be a useful step in the right direction. 

That would then lead to an AI, that is then built upon layers so that on top of loop gravity there might be another layer where everything is either Rat, Cat, or Cheese.  That is unless you are a rat; or are something else.  If you are a rat, then there is cheese, and there are mazes.  There is also sex, and there are the mazes that lead to sex.  There are also traps, and predators, and prey.  There might be locked doors, or other barriers to one’s progress … whether one desires wine, cheese, sex, or sleep – or all of the above!

Everything is cheese!  Everything is sex!  Everything is Mazes! Or else there is rock, scissors, paper!

Even if all I do is procure an acrylic fish tank, add a Parallax propeller and some blinking LEDs, so that different LEDs will light up according to different "emotional states".

There is also the notion of survival.