The big tech companies are now, according to some sources, collecting data with such a frenzy to win the race to be the first with a strong AI1. If we understand anything about the human condition, it is that the collection of knowledge through direct experience in interaction with the physical environment is central. What kind of AI do they think they can construct using large amounts of behavioural data? Do they see humans as just a collection of behavioural expression data and if it is collected within a beaker (aka. AI) it will automatically exhibit human characteristics? Predict: Yes. Mimic: Possibly. But is that the same as being? Mimicking a superficial expression of behaviour is not the same as creating an internal mechanism that gives rise to functionally similar, but not identical, expressions of intelligence. The latter type, the one that achieves similar goals, but through different routes and with different aesthetics, is what I’m after.
However, this is more likely just an alibi to increase their commercial predictive power unhindered (my note December 2019).↩︎