Episode #184 ... Is Artificial Intelligence really an existential risk?

Stephen West opens the episode with a thought-provoking question regarding the neutrality of technology. He references previous discussions on AI and consciousness, emphasizing that whether technology is 'good' or 'bad' often depends on its usage. He challenges the notion that technology is merely a tool, suggesting that some technologies may carry inherent moral implications.

The discussion transitions to what constitutes intelligence, with a focus on AI, particularly large language models like ChatGPT. West explains that intelligence isn't solely about human-like processes, arguing that AI can have its own form of intelligence. He outlines different categories of intelligence, highlighting narrow intelligence versus general intelligence and illustrating these concepts with examples such as AI in chess and image recognition.

As the conversation progresses, West explores the implications of creating a superintelligent AI. Drawing upon Sam Harris's thought experiments, he discusses how superintelligent entities may view humans and sets a stark contrast between human intelligence and the cognitive capabilities of machines. He warns that such entities could pursue goals that, while logical to them, might be disastrous for humanity.

West dives into the complexities surrounding AI alignment – ensuring an AI's goals match human values – and control, which involves keeping superintelligence contained. He discusses how naïve strategies to merely control an AI may be insufficient and emphasizes the philosophical debates around potential unintended consequences that could arise from simple goals programmed into highly intelligent systems.

The episode includes a dialogue format presenting skeptical views regarding fears of AGI. Skeptics argue that projecting human characteristics onto AI might lead to unnecessary fears. West, however, counters that many avenues of AI development could lead to unforeseen risks, necessitating ongoing conversations about mitigating those risks before they become realities.

West concludes by reiterating the primary argument that technology is not neutral; it carries specific affordances that impact society. He emphasizes the urgent need to discuss the ethical implications of emerging technologies, particularly AGI, to prevent potential societal harm.