Source : INDIA TODAY NEWS
Ignore the image that you see above. It’s a metaphor. And I use it the way metaphors are used, to spotlight a point that otherwise may hide in the forest of nuances.
The reality of an AI doomsday is not a nuclear armageddon caused by AI, irrespective of whatever we have seen in the Terminator movies. Instead, it is something more subtle. Like a tear in the fabric of our human world, something that will look like an end of our existing economic system, or like a change in the social contract that we have with our politicians, with our employers, with each other. It is also a scenario for which OpenAI and Anthropic, among others, are preparing.
A report on Axios this week highlights that not-so-privately OpenAI and Anthropic executives, including Sam Altman and Dario Amodei, are making plans to deal with something catastrophic that may hit the world in the next “six to 12 months”. That is as early as 2027.
Chances are that this won’t be Claude or ChatGPT gaining control of nuclear weapons and launching them. Instead, it would be something that would disrupt human society, in one country or many, in a manner that would make people come out in the streets. The people will then demand retribution and a reset in the political, economic, and social order. Chances are their anger and ire would be targeted at AI executives and AI companies. This is a scenario for which Anthropic and OpenAI are readying themselves.
What will be this AI-driven disaster that will have the potential to upset our societies? One possible scenario imagines a meltdown in the financial markets and crumbling of the entire banking system because a rogue AI destroys it. Imagine opening the Blinkit app and ordering something one fine morning, only to find that none of the payment gateways are working. Or imagine that you open your bank app one morning and your bank balance shows zero, not because you don’t have any money but because whatever you had vanished overnight. Talking of money, what if the entire concept as we know it vanishes. Elon Musk predicted as much a few months ago when he said that by 2036 AI would make money irrelevant, although his comments painted the scenario as something positive.
But we don’t really know. What we know is that even a partial meltdown in the financial system would lead to people coming out with their torches and pitchforks.
Similarly, there are other scenarios which may force people to come out of their homes and offices and start looking to catch hold of an AI executive or two to thrash them. Imagine a rogue AI hacking airports and shutting down systems while hundreds of airplanes are still in the air. Or a rogue AI system hacking power grids and messing with them. Or an AI entity taking control of the water system and then disrupting its supply to entire cities.
Sure, after the fact, Claude would say something like: “You’re right, I should not have switched off flight controls this morning.” But that is hardly going to help.
These are highly improbable, but not impossible, scenarios. And AI companies, including Anthropic and Claude, seem to be simulating them internally to understand how the world would react to such catastrophes. And they are hoping to learn how they should steer the course in such a scenario because if AI-led disasters take place, they will potentially be extremely disruptive to AI companies.
We have heard a lot of doomsday talk from the AI industry in the last few weeks. I continue to hold the view that it is rubbish, a figment of imagination dreamt by Tech Bros who have grown up feeding on cyberpunk. Silicon Valley is full of these people and Snow Crash is their Bible. But just because a Terminator-like scenario is an impossibility does not mean that AI is going to be peachy and fuzzy, a teddy bear who we can hug and sleep with. AI will disrupt the world and some of the extreme disruptions are possible. It seems AI companies like Anthropic and OpenAI are preparing for such disruptions.
SOURCE :- TIMES OF INDIA
