Is Sentient AI Possible? 8/29/23

In Saturn Conundrum, you watched Jason evolve from a ho-hum AI assistant to a fully sentient intelligence. In Saturn Rendezvous he takes on an android body and Human characteristics, becoming indistinguishable from a normal Human - the Holy Grail of AI.

Great grist for SciFi, but is it achievable in the real world? Until recently, I thought it was. With enough memory, processing power, and neural networking, surely it was just a matter of time before a computer could think and respond like the Human brain.

Curiously, ChatGPT, purportedly the cutting edge of AI and the poster child of machine learning programs, has caused me to revise my thinking. And not because modern AI programs sometimes give wrong answers, even garbage, in response to a user's queries. I believe these problems will eventually be ironed out.

What is missing with regard to sentience and consciousness is credible research into what these two attributes actually entail, both in the biological world and in computer technology. We have the 'how' and 'what', but are completely devoid of the 'why'. Until AI programs can evaluate queries and the data they produce in the broader context of their meaning and how they might apply to unrelated issues and unasked questions, sentient AI will remain beyond reach. For all the progress we've seen recently in AI, virtually no significant progress has been made to resolve this last issue.

For a more comprehensive account of this issue, I recommend The Myth of Artificial Intelligence, by Erik J. Larson.