Expert warns AI's real danger is premature trust, not superintelligence
Andrew Rogoyski argues that AI is unlikely to become a malevolent superintelligence, but rather the greatest risk is that society will trust it too soon and be disappointed. He suggests that AI will fail to meet exaggerated expectations, leading to a backlash rather than a Hollywood-style war.
Rogoyski's warning reframes the AI safety debate away from apocalyptic scenarios of machines seizing control. Instead, he identifies a more immediate and human-centered risk: the gap between marketing hype and actual AI capability. When systems fail to deliver on promised breakthroughs, public confidence could collapse, potentially stalling legitimate research and practical applications.
This perspective aligns with historical patterns in technology adoption, where inflated expectations often precede disillusionment. The article suggests that the cycle of hype and disappointment poses a greater threat to progress than any hypothetical superintelligence. Rogoyski's position implies that managing expectations may be as important as advancing the technology itself, particularly as AI becomes embedded in daily life and critical infrastructure.
If Rogoyski's assessment proves accurate, the primary affected parties could be businesses and institutions that invest heavily in AI based on current enthusiasm. A widespread trust collapse may slow adoption of genuinely useful applications, leaving organizations cautious about integration. Policymakers could face pressure to regulate prematurely, while researchers may struggle for funding amid skepticism. The broader public could swing from uncritical acceptance to undue rejection, complicating measured engagement with AI tools.