기후변화와 AI의 공통점 [윤지로의 인류세 관찰기]

In 1965, a report titled "Restoring Quality to Our Environment" was published in the White House, authored by the President's Science Advisory Committee. The environmental pollution report detailed multiple sources of contamination, with the fourth being carbon dioxide in the atmosphere. At that time, it was an unfamiliar pollutant.

The report warned that carbon dioxide accumulation in the atmosphere could lead to 25% more in 2000, potentially causing measurable and possibly significant climate change. Sixty years later, we are still grappling with the same issue. In 1990, the Intergovernmental Panel on Climate Change (IPCC) began reporting concerns that greenhouse gas emissions must be contained if we do not want them to continue increasing.

The warning signs have been consistent for decades, yet humanity continues to ignore them. The atmospheric carbon dioxide concentration, which was 320ppm in 1965, has risen to 430ppm today. While scientific understanding has improved and warnings have intensified, emissions have only continued to rise. Recently, a similar scenario has played out with artificial intelligence (AI).

Antropic researcher Alexei Tolstoy resigned from the company, stating that he sincerely believed AI could kill us all within 10 years, and the company's CEO argued that the speed at which AI models are being improved should be intentionally slowed down. AI leaders from OpenAI, Google DeepMind, and xAI echoed these sentiments. This has led to discussions about "pruning the growth" of AI, with concerns ranging from a "stop" command for AI development to accusations that big tech are using the hype to justify astronomical investment losses.

However, the point remains that AI warnings are not new, and we all feel the unprecedented speed of development. The question remains, can we afford to delay slowing down this rapid advancement, given the clear concerns about its potential consequences?