This TED talk by Peter Molyneux:

…demos Milo, a hotly anticipated video game for Microsoft’s Kinect controller. Perceptive and impressionable like a real 11-year-old, the virtual boy watches, listens and learns — recognizing and responding to you.

The demonstration begins with an explanation of how Milo is constructed. A combination of the following three elements allow Milo to exist:

1. A Kinect Camera
2. Artificial Intelligence developed by Microsoft
3. Emotional Artificial Intelligence built by Lionhead Studios.

Milo moves through a synthetic environment predicated on User-directed biofeedback/body gestures: no mechanical controllers are necessary. Unfortunately, Milo’s introductory learning curve [which is integral to the "game" leveling system] involves inherent gender bias: if you’re a girl, your initial game variable is a Butterfly whereas if your a boy, you’ll be presented with a Snail.

The demonstration goes on to illustrate how Milo’s face is comprehensively AI driven. His facial movements include blush response, nostril “flare” size [indicating stress], “body matching” [causing neuro-linguistically driven facial alterations] and responses to verbal cues. Peter then describes how Milo’s personality development is predicated on a Cause-and-Effect dynamic. This causality is showcased via 3 examples:

1. The User can choose to direct Milo to squash a snail: if the User does it will effect “…how Milo develops”.  The specifics of the verbal stimulus employed including how the User vocalises [specific phrases and intonations] all contribute to a database that informs and effects future interactions.
2. The User teaches Milo to skim stones over the surface of a river [skewed gender stereotyping is again evident here].
3. The User choosing to clean Milo’s room: Milo’s recognition of the User’s beneficial intervention and verbal engagement promotes sustained developmental interaction based on [what Peter terms] “deep psychology”.

This “deep psychology” [or what is described in synthaptic terms as "augmentology"] encourages a User’s empathy loadings. This in turn allows such games to shift towards complex experientially-defined engagement. These games surpass the hollow reinforcement of contemporary Social Games such as Farmville: instead, the User “levels up” by knitting fictionalised engagement with personality/identity construction and personalised growth variables. The element of cloud-directed learning [coaching synthetic humans whose social and chronological development depends on "crowdsourced" input] creates enormous opportunities for instruction and feedback via these types of  “Reality Gaming” systems [highlighted here by Seth Priebatsch]: