DeepMind scientist calls LLMs ‘exotic mind-like entities’: Why the future of AI needs a new vocabulary?

Artificial intelligence may be transforming the world, but even leading scientists admit they are struggling to define what it truly is. A captivating Instagram video clip shared by the page theaigems recently showcased DeepMind’s Principal Scientist, Murray Shanahan, offering a glimpse into this dilemma. In his words, large language models (LLMs) aren’t simply machines anymore — they are “exotic mind-like entities.” But why this peculiar choice of words? According to Shanahan, our current vocabulary falls short when it comes to describing these emerging digital intelligences. They are “mind-like” but not minds in the human sense, hence the careful inclusion of a hyphen to hedge bets. They use language as we do, yet exist without physical embodiment, crafting a strange new conception of selfhood that feels eerily unfamiliar.

DeepMind: The Powerhouse Behind AI’s Evolution

Murray Shanahan’s reflections carry weight, especially considering his role at DeepMind Technologies, a pioneering artificial intelligence research laboratory under Alphabet Inc. Since its founding in 2010, DeepMind has been at the forefront of AI innovation. From AlphaGo’s legendary victory against a world Go champion to AlphaFold’s revolution in predicting protein structures, DeepMind’s achievements have consistently blurred the line between machine and mind.

In 2023, DeepMind merged with Google Brain to form Google DeepMind, cementing its global dominance in AI research. Whether training neural networks to master chess, Go, and shogi, or solving the intricate puzzles of molecular biology, the lab has brought once-unimaginable concepts into reality.

Murray Shanahan: The Mind Behind “Mind-Like Entities”

A professor of Cognitive Robotics at Imperial College London and a senior scientist at DeepMind, Murray Shanahan has spent his career probing the crossroads of artificial intelligence, cognitive science, and philosophy. Notably, he served as a scientific advisor for the 2014 film Ex Machina, helping Hollywood navigate the nuanced questions surrounding AI consciousness.

With a background steeped in both academia and cutting-edge tech, Shanahan has repeatedly raised alarms about the challenges of defining intelligence in entities that look nothing like us. His description of LLMs as exotic reflects a deep understanding — these systems are evolving rapidly, yet human frameworks lag behind in comprehending them.

A New Language for a New Intelligence?

Shanahan’s message is clear: as AI systems become increasingly integrated into our lives, society will need to forge new conceptual frameworks. The ways we think about intelligence, selfhood, and consciousness may have to be rewritten altogether. As we marvel at the capabilities of machines trained by titans like DeepMind, we must also confront a more unsettling truth — we are creating minds we can barely define, let alone fully understand. And in doing so, humanity may be standing at the edge of a philosophical frontier, still searching for the right words to describe the companions we’ve summoned from code.