I’ve spent the last five weeks neck-deep in the AI risk rabbit hole. And my worldview has taken a beating. I used to think the coming AI wave would be bumpy, weird, scary and fucked up, but survivable. Now I’m not so sure.
The deeper I dig, the more it feels like we’re barreling toward a cliff.
And I hate it.
Here’s the problem: most conversations about AI risk are all hand waving and gut feels. “People have always come out on top.” “We’ll adapt.” “Someone smart will figure it out.” Or they’re politicized “they’re just scaring us so winners get legislated.”
I needed something concrete to measure if I should actually be concerned, or if I could wind back the clock to before AI risk gnawed at my brain.
I couldn’t find a tool that did this in a way that felt realistic and customizable. So I built one. The AI Risk Calculator takes your assumptions – about when AI becomes smarter than us, how fast it improves, whether it can deceive, whether we stay in control – and multiplies them out to give you a number.
The number you’ll get is the chance that AI could do Very Bad Things that humans couldn’t stop. Not necessarily because it’s evil, but just because we’re in the way.
In the process of building this I educated myself on AI, and I’ve included citations in the calculator so you can learn about it too. You can tweak every slider and gate, see exactly how your beliefs about AI shape the risk.
I didn’t like building this. The idea of existential risk is scary; it reminds me of how I felt during my Cold War childhood. But if the odds are even remotely close to “ride off a cliff,” we shouldn’t just hide under the blankets hoping it will all be OK.
I’ll be writing more soon about what this means for us as a species, and how we might respond. But for now, try the calculator yourself. See where your own beliefs land you.
When I write next, I’ll share the AI risk score that I got and we can compare.
Try it out: https://airisk.rajjha.com
