On Thinking Machines

4 days ago

The article discusses the application of LLMs and AI-augmented tools, focusing on intelligence and consciousness.
Intelligence in AI is seen as an analogy to human or animal intelligence, not necessarily more or less intelligent than traditional computing.
Historical figures like Charles Babbage and George Boole contributed to automating intelligence through symbolic forms and logic.
Current AI tools like CNNs and LLMs are different analogies to intelligence but not inherently more or less intelligent.
Consciousness in AI is questioned, with the author believing neither traditional software nor LLMs are conscious.
Marketing strategies sometimes suggest LLMs might be conscious to evoke wonder, but the author dismisses this as exploitative if taken seriously.
The Three Laws of Robotics are critiqued as creating a dystopian, slave-like existence for conscious machines.
Consciousness lacks a broadly accepted test, making it difficult to determine if a computational system is truly conscious.
LLMs, despite their capabilities, are seen as deterministic applications of routine, lacking true understanding or goal-setting.

Hasty Briefsbeta