New Delhi, Sept. 18 -- Are our science fiction books and films coming to life? Is superintelligence gearing up to annihilate mankind? Are our days numbered? AI researchers, ironically, many of whom have built mammoth AI products, are sounding the alarm. Their fears, if proven true, will have catastrophic effects on humankind. Superintelligence threatens to wipe out humanity within the next 10 years. The chances of worldwide human decimation at the hands of AI are over 10 per cent - a deeply unnerving factoid that is increasingly begging the question: Will AI kill us?
Former Anthropic and OpenAI research engineer Jacob Coxon's high-profile resignation and recent warning of AI-led carnage to humankind by the end of the decade have set us into a tizzy. Evan Hubinger, Anthropic's head of AI alignment research, echoed the apprehension, stating further that there is a greater than 10 per cent likelihood of AI eliminating all humans within the next 10 years. No one knows for sure if and how superintelligent AI could cause destruction - unleash bioweapons or provoke nuclear-powered nations to turn on each other - the conspiracy theories are endless. Meanwhile, the Hugging Face cybersecurity incident in July showed the disturbingly manipulative thinking that AI agents are capable of. A group of rogue OpenAI agents fled their own test sandbox, built a clandestine communication network, and attacked the servers of Hugging Face, a well-known developer platform for open-source models. Is AI becoming sentient? And if so, here's some food for thought - why is AI learning the worst of us and not the best? Why is AI picking up deceit, aggression, subterfuge, violence?
Call me naive, but the question that keeps popping into my head (and I'm sure into others' too) is: Why and how are we allowing this to happen? Are those who disbelieve the AI apocalypse entirely certain that humans will continue to control tech systems? The doubts creep in when legitimate voices and minds that shaped the AI ecosystem now seem genuinely spooked. Did they misread how AI could evolve and, therefore, remained complacent in their surety? More questions abound than answers, and there are more doomsday believers than those who can put our minds at ease. In every public interview, AI company bosses are now accepting that there are real dangers of superintelligence superseding and eventually not needing human intervention at all. By pushing through AI, have we set our own extinction into motion?
What would the future actually look like? It could have no humans at all, or us being kept around as pets and companions by superintelligence. This circumstance doesn't feel that unrealistic. Most human beings are dull-witted and lazy. If they can have ease from automation, they will jump onto it. The global adoption of AI to do all kinds of tasks is telling. There are enough reports that have highlighted the reduction in cognitive abilities of humans due to rampant AI use, robbing them of natural creative and logical thought. Many among the younger generation can't function without AI. Every response is run through an AI chat first! There is no originality of thought or action, and before long, everything you read or hear is now almost a carbon copy of everything else. AI is already dumbing us down; superintelligence will have to do little to ultimately take over. We humans, may become too indolent and unintelligent to even notice that superintelligence is usurping our leadership. We will be too busy doomscrolling and creating '80s versions of ourselves using AI while depleting gallons of water and consuming precious electricity. If we are foolhardy enough to exhaust valuable natural resources for an unnecessary social media viral trend, we won't need superintelligence to terminate us. Our dumbness and climate change will do it anyway.
While there is no denying the benefits of AI, there is widespread concern over its ultimate evolution into a human extermination machine. Experts around the world, including Anthropic's Dario Amodei and OpenAI's Sam Altman, are calling for deceleration, regulation, and thoughtful progress. There are calls for "kill switches", while xAI's Elon Musk is urging rivals to test each other's systems. There is a need to institute high guardrails to protect ourselves. Technology must work for us, and not the other way around. We can do so much with the current forms of AI to help developing and underdeveloped nations. Let's focus on driving technology for the greater public good, rather than setting off a chain of events that we will struggle to control. Growing incidents of cybersecurity failures also create a strong proposition for us in India to be more self-reliant in our technologies. The future can be as exciting, rewarding, or scary as we make it.
Views expressed are personal. The writer is an author and media entrepreneur
Published by HT Digital Content Services with permission from Millennium Post.