"Anyone who engages in serious dialogue with a Large Language Model (LLM) may get the impression they are interacting with an intelligence. But many experts in the field argue the impression is just that. In philosopher Daniel Dennett's words, such systems display " competence without comprehension ". The hype about Artificial General Intelligence (AGI) from big corporations and their celebrity spokespersons has prompted a backlash, in which scepticism turns to cynicism, tinged with paranoia about how " stochastic parrots " may start to control our lives."
"Research on intelligence has a chequered history, tainted by eugenics, statistical manipulation and a banal obsession with metrics. Agüera y Arcas counters this by opening up the topic as wide as it can go. A physics graduate with a background in computational neuroscience, he is something of a polymath. He draws explanatory frameworks from microbiology, philosophy, linguistics, cybernetics, neuroscience and industrial history."
Large language models can create the strong impression of being intelligent, yet that impression can be misleading. The phenomenon is captured by Daniel Dennett's phrase " competence without comprehension ". Corporate hype around artificial general intelligence has provoked backlash, with scepticism often hardening into cynicism and paranoia about " stochastic parrots " influencing lives. Intelligence has become an overheated topic that requires calmer, less assertive thinking and a refreshed starting point. A computational framing treats computation as a tool to think with and suggests deep structures of reality may be computational. Research on intelligence bears a chequered history tainted by eugenics, statistical manipulation, and an obsession with metrics, and an interdisciplinary approach draws on microbiology, philosophy, linguistics, cybernetics, neuroscience, and industrial history to widen the inquiry.
Read at The Conversation
Unable to calculate read time
Collection
[
|
...
]