Two of the founders (2010) of Deep Mind (bought by Google 2014) had studied the neuroscience (the brain) and were avid gamers. Unlike IBM’S Deep Blue or Watson, which were developed for a pre-defined purpose, these guys were shooting for a system that was not pre-programmed: it learns from experience - like humans. The experiences they started training it on were games (they were gamers!) - early ones like ‘pong’ but getting increasingly sophisticated. Move along a few years and you get to the A.I. AlphaGo which beats the Go world champion - ten years before such a feat was considered possible.
The relevant issue here is that they were taking a new path in A.I. not just in the programming/training but with the architecture - they were developing a system using the ‘neurons, synapses, weights, biases, and functions'.[141] - components/processes that mimic the workings of the human brain.
Step away from the (currently obsessive) idea of A.I. as an actor for a moment and focus on an A.I. as reverse engineering of the human brain.
There is a huge spectrum of chemicals (alcohol, heroin, LSD, etc.) that can stimulate the brain in very specific ways so a person can feel (or not feel) certain things while those chemicals are in their system. These chemicals can also have long-term effects (often degenerative but not necessarily) both on the user and their fetus if they’re pregnant (see the final scenes of Contact-High Pilot).
If we have the architecture of the human brain we have a experimental model for how to stimulate it for very specific responses. Hypothetically that stimulation could be targeted in very specific ways using electromagnetic impulses rather than drugs.
What would the electromagnetic-impulse interface be?
NIBS or Non-Invasive Transcranial Brain Stimulation has been practiced for almost 4 decades using a variety of techniques such as Repetitive Transcranial Magnetic Stimulation and Transcranial Electrical Stimulation. The research has been focused on therapeutic options for psychiatric and neurological disorders and establishing protocols around permanent outcomes (synaptic plasticity) which might be negative - ECT (Electroconvulsive Therapy, in use for over 100 years) is still controversial and casts a long shadow over this research.
There are a number of highly sensitive ‘contact pads’ that can be used for stimulating the brain with electromagnetic pulses - though they much more in use for measuring electrical impulses from the brain than to the brain. Up until now a lot of NIBS research used animal models with post-experimental dissection. What if - like almost every other branch of scientific/engineering research - they worked with an utterly comprehensive architecture in a computer brain-simulation model? One that could identify specific responses to a variety of simulations.
So the final piece of this puzzle is the utter specificity of the Transcranial stimulation required for different experiences: the nano-timing and mix of magnetic vs electro pulses, amperage, voltage etc. What mix gets you the heroin high rather than the Viagra experience?
A contact pad must be flexible for maximum surface connectivity. Plus its contact interface must contain a massive number of ‘nodes’ (contact points) that can fire independently, or synchronized in tandem with other nodes, to produce highly targeted input to the brain tailored to the required experience. This technology could be adapted from GPU or similar chips with their nano-thinness (for flexibility) and staggering dense electromagnetic activity across a nano-thin layer.
So buy the contact-pad, shave the base of your skull and stick in on, wired to your phone (or maybe risk Bluetooth) go to the app store and download your chosen but limited and time sensitive experience.