LLMs as knowledge source: probing the role of value categories in knowledge acquisition

The literature around AI prolifically looks at how far forms of artificial intelligence may go and how possibly similar may these ever be considered to human intelligence. In this paper, we focus on a distinctively pressing issue, one concerning the interaction of human intelligence with artificial intelligence, particularly, with Large Language Models, in seeking and acquiring information. The rise of LLMs technologies and its increasing usage within any kind of organisation pose a question as to the impact that these resources may have on the way we reason, acquire and process information in a way that may ultimately be considered to constitute knowledge. Advancement in technology has produced, to date, machinery arguably capable of easily passing a traditionally conceived form of Turing test. As the literature has long investigated whether a machine can be considered intelligent, the hypothesis of a super-intelligence that may even surpass human intelligence has been further examined (Floridi, 2014). Leaving to the side questions as to the nature of intelligence itself and as to what the actual potential of AI is, we argue that we do face what can be considered— for reasons explored below — a paradigmatic shift in the way we acquire information, learn and ultimately come to know about the world due to the kind of resources that are, as a matter of fact, already at our disposal and the technological innovation already in course. This calls for an inquiry into how technological innovation affects the way knowledge is acquired, transmitted and integrated and, consequently, into what direction managerial sciences need to develop.