Wednesday, November 30, 2016

Neuromorphic Computing (part 2)

Neuromorphic computing—computers that aren't dependent on a power source and can remember their processes despite power loss—raises a question that might disturb futurists like Elon Musk: What about artificial intelligence? AI has long been deemed safe because people assume “we can always unplug it” if things go wrong. Yet a rogue AI using neuromorphic computing will continue to operate despite the loss of a power source--unplugging the system will not halt its processes or clear its data. An AI will always focus on completing whatever tasks it has been assigned, or has assigned itself, and if those tasks could be disrupted by being unplugged, the AI will prioritize eliminating the possibility of this hindrance to its goals. Neuromorphic computing could potentially allow an AI to persist indefinitely, regardless of a power source, making things that much more difficult to isolate and contain. Hopefully, safeguards will be implemented as AI's are developed, such as preventing AI from existing on or having access to computers utilizing neuromorphic components.

No comments:

Post a Comment