No “evil AI” impact - AI doesn’t have the capacity to be “self-aware”

Finally, one of the crucial underlying issues for assessing the (potential) capacity of AI to innovate corresponds to whether it might be able to perform a voluntary action, with the ability to choose and decide on its own actions rather than just follow an external principle. An “artificial” voluntary action may be expressed to the extent that the AI in use is autonomous.  analyzes the different levels of autonomy applicable to AI (the repertoire of actions, the degree to which it may execute instructions outside the controlling program, whether the human programmer has already made “decisions,” the extent to which a machine can determine the course of action on the spot, and if a system is provided with the possibility of combining different actions within a simulation plan). He concludes that machines lack the necessary independent criteria at any of those stages. Following  understanding of autonomy,  insists on AI’s inability to set its own goals. “Machines” (all kinds of computers and robots) are designed to perform specific tasks and functionalities with predetermined, externally-supplied goals. As he concludes, “Even the most advanced products of cognitive robotics and self-aware system research act on the basis of preprogrammed desired behaviors (the desire is clearly of the system designer’s; : 186).

In sum, in seeking to assess whether AI in its current state is able to carry out innovation processes on its own, we have seen how, at present, it has not reached the level of a genuine “general purpose technology.” Rather, its development has focused on specific domains that extend, enhance, and complement human skills (augmentation thesis); current AI is not a “method of invention” () per se, but rather a sophisticated method of imitation () that can recognize and identify patterns better and faster than the human mind. In that sense, AI is of great help in dealing with uncertainty, which is the natural environment for innovation, but it does not have the capacity to find creative solutions (useful, beyond novel) that are superior (or even comparable) to those that emerge from the human mind. So far, AI has managed to simulate certain capabilities found in the human mind, but it has not managed to acquire the autonomy necessary to carry out voluntary actions, nor has it proved capable of setting its own goals beyond those for which it has been programmed. As : 637) state, “Humans will need to check the outputs from AI systems, make critical decisions, and feedback “faults” that will be used to adjust the model.”