Projects

Peer-assisted learning (PAL) and hybrid collective intelligence

Research efforts in Psychology have traditionally favored learning paradigms that emphasize incremental skill refinement and enhancements in fluid intelligence. In contrast, research on knowledge-based, crystalized, collective, and hybrid human-machine intelligence has not received sufficient attention and resources, definitely not on par with its potential. The impact of the latter can be significantly greater than the former. In certain conditions, a group of interacting intelligent human and artificial agents can become a powerful knowledge multiplier that has the potential to enhance our ability to solve problems of increasing complexity, such as the climate change problem.

Under the assumptions that knowledge is distributed in the population of learners and different learners have different learning experiences, interaction among learners provides opportunities for exchanging knowledge and filling the gaps in learners’ minds (an effect known as knowledge spillover). However, learning from other learners has risks as well. For instance, individual learners may have incomplete, erroneous, or biased knowledge, which may be compounded by interaction among learners. In addition, learners may become too reliant on other learners and less inclined to exert sufficient effort individually to develop and maintain their knowledge base. In the PAL project, we study both the opportunities and challenges of peer-assisted learning when peers are either humans or intelligent agents. The ultimate goal is to facilitate human learning and augment collective intelligence.