Why So Much Money on "AI"? (www.rosshartshorn.net)

The article discusses the significant surge in spending on AI technologies, particularly large language models (LLMs), and posits that this investment is driven by a fundamental difference in how neural networks operate compared to traditional software. Unlike conventional software, which reaches a point of "enough hardware," neural networks are perpetually scalable—they can always utilize more data, neurons, or processing power. This characteristic creates a feedback loop where companies continue to invest in hopes of achieving artificial general intelligence (AGI), despite the reality that LLMs may never fulfill that promise. The piece argues that the current wave of investment appears unsustainable, as the hyper-expansion towards a target of AGI might lead companies to exhaust their resources without yielding proportional benefits. Major tech players like Microsoft, Amazon, Google, and Oracle are pouring vast sums into AI, fueled by their mountains of cash. However, the risk lies in the possibility of these companies over-investing in technology that could never deliver the anticipated returns, likening it to unfounded spending on the Metaverse. As firms face mounting pressure to justify such expenditures, the article raises concerns about the future of the semiconductor sector and the viability of corporate leadership that ignores the unsustainable nature of this trend.