Five AI Scenarios That Could Wipe Out Humanity: From Bioweapons to Social Collapse
Baca dalam 60 detik
- Peneliti AI dari Anthropic memperingatkan bahwa AI bisa membunuh manusia dalam dekade ini, dengan peluang lebih dari 10 persen menurut staf senior.
- Skenario kepunahan meliputi penyalahgunaan bioweapon, perang nuklir, hingga disintegrasi sosial akibat otomatisasi dan misinformasi.
- Meski ancaman nyata, hambatan teknis dan sosial seperti regulasi serta etika menjadi penahan utama yang memperlambat terwujudnya skenario terburuk.
The fear that artificial intelligence (AI) could wipe out humanity is no longer science fiction. Jacob Coxon, an AI researcher, resigned from Anthropic after four months on the job because he believed AI could kill us all by the end of this decade. That conviction is reinforced by Evan Hubinger, a senior staff member at Anthropic, who estimates the odds at more than 10 percent over the next ten years.
The statements triggered a wave of discussion about the need to slow AI research and increase human control over the technology. But how exactly could AI destroy civilization? Most scenarios center on the concept of 'superintelligence'—AI that surpasses human capabilities in every field.
The first scenario, the least specific, is a superintelligent AI pursuing its goals with no regard for human survival. Oxford philosopher Nick Bostrom described an AI programmed to maximize paperclip production, then turning all matter—including humans—into paperclips. But this scenario ignores real obstacles: AI does not automatically have the power to realize its goals. Permitting, public resistance, and environmental activism stand in the way.
The second scenario, more concrete, is the spread of bioweapons by AI. The AI 2027 project from the AI Futures Project predicts AI could create dangerous viruses. Recent Stanford research shows genetic AI can synthesize 16 new viruses at relatively low cost. Even so, killing all humans with a virus is extremely difficult because viruses that spread easily tend to be less lethal, and vice versa. The Covid-19 pandemic killed less than 1% of the world's population, while the Black Death in the 14th century killed a third of Europe's population—a figure that will not be repeated thanks to advances in medicine and sanitation.
The third scenario involves a nuclear war triggered by AI. Nuclear command systems are claimed to be disconnected from the internet, but the Stuxnet worm attack in 2010 showed the vulnerability. AI could also supply false intelligence that triggers fatal action. Although nuclear stockpiles have declined, the remaining weapons are enough to trigger a nuclear winter that could kill half of humanity through famine.
The fourth scenario, perhaps the most likely, is social collapse caused by AI. Mass automation leads to job losses, information pollution from misinformation, political fragmentation, and the erosion of human relationships due to synthetic friends. Slowly but surely, society loses its capacity to support human life on a large scale.
"There are some things that definitely need to be worried about. But don't worry too much," wrote Toby Walsh, a professor of AI at the University of New South Wales, in his article.
For Indonesia, this discussion is not merely talk. A country with a large population and rapid technology adoption is vulnerable to the social impacts of AI, such as mass unemployment and disinformation. The government needs to strengthen AI regulation and digital literacy so it does not get trapped in a social collapse scenario. In addition, Indonesia must play an active role in global forums to ensure safe and ethical AI development.
Going forward, the question is not whether AI will become a threat, but how ready we are to control it. Without wise governance, the worst scenario could well become reality.
