Context: Pentagon Wants AI to Predict Events Before They Occur
If we abstract the concept of pattern formation from its instances, this becomes plausible. This becomes about not specifically determining what is going to happen but rather and as a function of the resonant (i.e. harmonic) symmetries endemic of and to complex systems, where and when something interesting as salience of novelty is likely to occur. Semantic interpretation as a sophisticated sentiment analysis-type technology would be a post-processing of inferred patterns and contexts in which they arise. When interesting things percolate through the system, assign them to real (i.e. human) brains for triage and consideration.
All of this requires an unorthodox approach to measurement and salience mapping that I expect might be (inadvertently) antithetical to the institutional context in which it is being developed. The axioms of measurement and determination of “interesting” patterns are above and beyond simple teleologies endemic of our own language and affiliated or derived cognition. Even as technology (and science) marches forwards, most of our conceptual vocabularies are still stuck somewhere in the mid-19th Century.
Large events do not occur in isolation and generally inflect self-gravitation as a function of complexity (and vice versa).