Photo by Ashithosh U on Pexels

The roots of modern artificial intelligence may be more feathered than you think. While pioneers like Alan Turing are celebrated for their theoretical contributions, the practical foundation for reinforcement learning, a crucial element of AI, can be traced back to B.F. Skinner’s mid-20th century experiments with pigeons. Initially aimed at a rather unusual wartime application – guiding missiles – Skinner’s research demonstrated the remarkable capacity of pigeons for associative learning, linking actions to rewards through trial and error.

These principles, once dismissed by some psychologists, were later embraced by computer scientists, paving the way for reinforcement learning, now a cornerstone of AI. Turing Award winners Richard Sutton and Andrew Barto are among those who have successfully applied these principles to enable computers to master complex tasks like autonomous driving and the board game Go. The success suggests that AI’s capabilities are built upon the ‘lowly principles’ of associative learning rather than a perfect replication of human intelligence.

AI’s achievements are now sparking a renewed interest in animal intelligence. The ‘associative learning paradox’ underscores how a process once considered too simple to explain complex animal behavior is now celebrated for enabling human-like performance in AI systems. Researchers are actively re-evaluating the role of associative learning in intelligent animals, recognizing the potential complexity of creatures long perceived as simple.

Johan Lind of Stockholm University emphasizes the importance of human symbolic language and cumulative cultures in distinguishing us from other species. Meanwhile, research led by Ed Wasserman at the University of Iowa showed that pigeons can detect cancerous tissue in medical scans with comparable accuracy to experienced doctors. These findings prompt comparative psychologists and animal researchers to reconsider how we attribute sentience to other living beings, especially in the context of AI’s rapid advancements.

While AI and animals may share associative mechanisms, ethical considerations are paramount. A pigeon can experience pain and suffering, while an AI chatbot currently cannot. As noted in Aeon, the surge in AI research investment compels us to address the possibility of AI sentience, highlighting the critical need for parallel investments in research on animal cognition and behavior. From Cold War missile guidance systems to cutting-edge AI, the laboratory pigeon’s legacy continues to influence both our computers and our understanding of intelligence itself.