AI is getting cheaper more quickly than any other transformative tech in history (epoch.ai)

Recent analysis reveals that AI performance costs are plummeting at an unprecedented rate, dropping approximately 47% per quarter. This remarkable decline translates to a staggering 13-fold reduction annually, outpacing any transformative technology in history. The study, which evaluated data from various AI benchmarks—spanning mathematics, hard sciences, and game-based puzzles—indicates faster cost reductions in mathematical tasks (50-52% per quarter) compared to game challenges (39-43% per quarter). This trend has likely been accelerating since the release of OpenAI's GPT-3 in late 2021. The implications of this cost drop are significant for the AI and machine learning communities, as they may lead to wider accessibility and increased applications of AI technology across various industries. Despite these falling prices, overall AI spending may remain robust, particularly for complex tasks that require high cognitive processing, such as reviewing extensive scientific literature. The research highlights a key paradox: as AI becomes more affordable, its capabilities continue to advance, establishing large language models as a pivotal "general-purpose technology" likely to influence multiple sectors, similar to the steam engine or electricity. This rapid evolution in AI performance and cost dynamics promises to reshape how technology is leveraged in both current and future applications.