Information abounds on how to prepare for earthquake or similar natural disasters: have pre-packed overnight bags, store extra water on premises, etc. Investing in a few small efforts can result in disproportionately positive outcomes should disaster seek you out.

Can you do the same thing with regards to potential world-ending AI doom scenarios? Timelines get shorter every day, advances are coming faster and faster, and it would seem – dare I say it – rational to actually make personal life changes to prepare your household for what is likely to come.

I have done some initial thinking on this myself, but wanted to provide at least a starting point for ideas and conversation. So first, a few thoughts on standard preparations:

Some preparation is disaster-specific. A tornado safe-room in the cellar is not useful during flooding. Unfortunately, we can't predict what AI FOOM looks like on a personal level, so this category of preparation is difficult to perform.
Some preparation is not disaster-specific and comes from human needs: food, water, shelter. There may be little to no preparation you can do on this front that will matter post-singularity, but the preparation is simple, low-effort, and might be useful. Think standard "prepping" things: stocking fresh water, food storage, blankets, flashlights.

Beyond this is where my certainty – and perhaps committment – wavers.

Financial Planning

Up until this point in my life I've followed standard advice for financial planning. I've been well to-do, and planned to retire early with significant savings. Even so, retirement is 10+ years out. If I discovered today that I had terminal cancer, there are many changes I would make to my lifestyle:

Quit my job immediately
Maximize time spent with family and friends
Significantly increase spending on experiences
Potentially sell my house (mentioned as an example of the hugeness of the potential life changes, but I won't go into specifics of why on this one just to keep things simple)

Well I haven't been diagnosed with cancer, but I do assign not insignificant >5% odds to AI FOOM within the next decades. Shouldn't I be applying the above points in some degree? Unfortunately, some of them are large binary choices that are hard to go back on later, and which I would regret if I were wrong. And if I do nothing, I risk squandering the short remaining time left to me.

Some half-baked ideas ...

Posts

Wikitag Contributions

Comments