Artificial Intelligence as Expected Intelligence

The inquiry into the nature of intelligence within artificial intelli- gence (AI) has been a persistent pursuit since the inception of the discipline, notably evident in Turing’s seminal works predating the formalization of AI itself. Turing sought to establish the viability of thinking machines, laying the groundwork for subsequent reflections on attributing intelligence to ar- tificial artifacts, particularly in the form of software programs. This chapter deals with the issue of measuring intelligence in artificial artifacts, empha- sizing the importance of considering the expected intelligence. From an interactive standpoint, the human expectation of intelligence in an artificial entity revolves around performances deemed cognitively suitable for mean- ingful interactions. This perspective advocates for the detection of varying degrees of intelligence that align with the requirements of reliable and cog- nitively relevant interactions, even at intermediate levels. The discussion delves into the inherent dierences between the simulative nature of human intelligence in artificial artifacts and the distinct characteristics expected for genuine intelligence in interactive contexts, particularly within the realm of social robotics. Indeed, the analysis of specific aspects of intelligence in robotic artifacts may prove enlightening by showing how the measurement of intelligence is significantly contingent upon the expectations that the human participant retains during interaction with AI. The inherently collaborative and interactive nature of the relationship between humans and robots un- derscores particular dimensions within the analysis of intelligence that might remain latent in interactions with disembodied AI systems.