Meet our new PIs: Simone Parisi on intelligent exploration, autonomy, and AI awareness

Simone Parisi first became captivated by reinforcement learning (RL) during his master's degree. It is a fascination that led him to pursue a doctorate and has since shaped his entire career. "Reinforcement learning is a distinctly human-inspired approach to machine learning and artificial intelligence, in which we seek to replicate the adaptive learning behaviours of humans," says Parisi. "It is a field defined by deep challenges, yet one that offers intellectual reward when those challenges are overcome."

Central to Parisi's research agenda is the problem of intelligent exploration: how machines can expand their capabilities and behave curiously without continuous human guidance. "My goal is to develop intelligent agents capable of mimicking human behavior, that explore out of intrinsic curiosity and operate autonomously even when external feedback is unavailable." A key part of this vision involves giving AI agents transferable, general-purpose capabilities, so that they don’t need to re-learn behaviors across different environments or fundamental skills across similar tasks.

Reflecting on his new professional home, Parisi speaks warmly of the environment in Finland. "Tampere University offers an exceptional research environment for AI and machine learning, and the ELLIS Institute provides a first-rate network and the resources needed to pursue my research. Beyond academic excellence, the genuine sense of welcome I encountered was a decisive factor in my decision to join."

Parisi is currently assembling his research group and is clear about the qualities he values: strong communication, intellectual perseverance and curiosity, and deep understanding of the mathematical foundations underpinning AI systems. His advice to prospective doctoral students is simple: "Maintain a critical perspective on your work, conduct research that fascinates you and creates value for others, and cultivate the ability to collaborate with peers." In the context of reinforcement learning specifically, he welcomes researchers who want to "bridge the gap between theoretical RL and the complexity of practical applications".

While the current wave of AI development is making these technologies more accessible and broadening their societal impact, Parisi cautions that the potential for misuse must not be overlooked. "We have a responsibility to reflect critically on the pace and direction of progress. My hope is that the research community can help shape public discourse, encouraging people to consider not merely what AI can do, but how to use it responsibly.” 

Parisi’s outlook on the future is upbeat. "I am genuinely excited about the work ahead at the ELLIS Institute, both the contributions I hope to make alongside these outstanding colleagues, and those that will emerge from the team I am building."