#wily devnotes

The more processing power an AI gets, the more complicated the situations it can handle at theoretical maximum load. By the same token, it will also be able to handle non-maximum situations more effectively, resulting in unused thought-time. It is clear that the trend of Artificial Intelligence is going to continue to be closely married to the idea of basing itself off the patterns of the human mind.

Building on ideas presented in my previous paper, if the field continues in this manner, it is entirely possible that a sort of truly independent thought could be within striking distance after enough progress. It could, theoretically, be possible to advance to a point where the resulting AI could behave in a convincingly human manner in most situations, complete with personality and conscience.

While this may still seem far-fetched, it still is a viable destination, and indeed, one that almost seems inevitable as a goal at some point. A potential pitfall to watch out for is simply: as far as one might aim to mimic the human mind, a robotic brain still works via different processes, and will not be predictable by the same models of human behavior in all situations. Especially true in cases where a human would already be unpredictable.

The human mind, over time, has developed a large number of strategies to preserve normal and safe functioning. Coping mechanisms that, to varying degrees, are there to prevent past events or biological/mental ‘malfunctions’ from causing damage to the person. These, already in the case of humans, are not entirely reliable: humans continue to be potentially stricken by a range of conditions despite both inbuilt preventative measures and modern medicine, which can alleviate some but not all of the cases. A robotic mind, especially based as close as can be on a human way of thinking, could still fall prey to such emotional problems if so equipped (as well as possible problems all on their own owing to the difference in natures)

The human mind is an intricate thing, a balancing act of countless distractions and drives. As robotics and artificial intelligence is pushed farther, the tendency has been to use the human mind as a basic outline of where to start. This approach is not without drawbacks, but those are outside the scope of this paper. Instead, the focus will be on pitfalls that will need to be avoided with the approach as technology progresses.

A simple AI could easily be viewed as a math equation, where putting something in one end spits out a result on the other side. Simple is potentially a problem though, as with bad input, you’ll get bad output. Adding complexity to check against errors becomes more work the more robust you want the correction to be. Adding the ability to handle new forms of information, to go beyond just an error-checking equation to a learning entity, is part of the holy grail of this field.

The focus thus far has been specifically focused on critical thinking patterns and information learning, with personality development being largely a byproduct and otherwise not a main focus. Strictly speaking, there are many cases where a ‘thinking’ machine would be more efficient than a ‘thinking and feeling machine’, but in the long run the technology will continue to advance, perhaps even leading to a point where robots are valued for more than just the work they can do easier that humans.

Time and the elements had worn away almost any trace of writing on this paper, which wouldn’t have drawn much attention in the remnants of a lab full of half-functioning computers and an experimental advanced robot that had killed most of the team that found it.

The handwriting, if it could be more readily made out, is frantic at first, disorganized. “What if a robot could be made that could do more than just think like a human? Aside from the advanced capabilities, it could be possible to fake basic human functions, allowing blending in. Eating? Recreation? What if nobody but the robot can tell it is a robot?”

--thinks that by trapping his intelligence in an endless cycle of tests and lessons he can make sure that his new

While what I’ve seen of the design does suggest very enhanced capabilities, even more than our current technology, I also know from the past that even a few hours of Light lecturing tests human endurance and patience. Decades of his doddering lessons and tests are more likely to stunt development. I would half expect the result to have a thin veneer of composure that immediately shatters under pressure under stress.

Such a waste of potential, to make a new mind and then hamstring it like that. Far better to --