How to survive until AGI

Some sources of personal risk are easy to avoid.

In my previous post, I made the case that surviving until AGI seems very worthwhile, and that people should consider taking actions to make that more likely. This post goes into what the most low-hanging fruit are for surviving until AGI. I’ll assume that AGI is less than 20 years away.

I’ll rank the various interventions by how many micromorts they reduce (1 micromort = 1 in a million chance of dying). The optimal strategy varies a bit based on how old one is because the older you get, the more of your micromorts are chronic rather than acute.

Main anti-recommendations

Hard drugs (tens of thousands of micromorts a year)

If you’re a young person, by far the easiest way to reduce your risk of dying in the next 20 years is to never take hard drugs. Using heroin costs around 30 micromorts per trip, using MDMA costs around 13 micromorts per trip¹, and a daily use of heroin has a mortality rate of around 2% annually (which is 20 thousand micromorts). Trying any hard drug is probably a very bad idea, as it runs the risk of turning into a very dangerous addiction.

Mountaineering (hundreds of micromorts per ascent)

The higher mountains seem much more dangerous to climb than others, with the micromorts of ascending Mt. Everest in the tens of thousands. For more typical mountaineering, Vasco Grilo estimated the risk at around 200 micromorts per activity.

Extreme sports and snow sports (tens of micromorts per activity)

Some sports are more dangerous than others:

• Paragliding: 74 micromorts per launch

• Skydiving: 8 micromorts per jump

• Rock climbing: 3 micromorts per climb²

• Skiing: 0.7 micromorts per day

Motorcycle rides and long car rides (tens of micromorts per trip)

Driving a motorcycle costs 1 micromort every 6 miles, meaning any meaningful distance covered by motorcycle will cost many micromorts. Cars are more than 10x safer, at 1 micromort per 250 miles. At that rate, you only start incurring large numbers of micromorts on multi-hour drives. For long trips, you’d be safer substituting car travel with a plane ride, which costs 1 micromort per 1000 miles, or even better, a train ride, which costs 1 micromort per 6000 miles.

Not wearing a seatbelt (tens of micromorts a year)

Not wearing a seatbelt roughly doubles the per-mile micromorts of driving. Assuming you drive around 250 miles on an average week, you’d be costing yourself a micromort per week, or around 50 micromorts a year.

Distracted or drunk driving (around one micromort per 25-mile drive)

I couldn’t find a source for the micromorts of drunk driving, but my best guess is that the per-mile risk of driving while slightly drunk is around 10x the risk of sober driving. So a 25-mile drive while slightly drunk will incur around one micromort. But one thing that’s important to note is that this is the only activity on the entire list that also incurs micromorts on other people.

Driving while distracted (texting, and to some extent talking) is probably pretty similar in terms of risk.

Chronic micromorts

Smoking and drinking cost notable amounts of chronic micromorts–around one per 1.4 cigarettes, or around one per 0.5 liters of wine. However, it’s unclear how much of that risk is incurred before AGI (when it matters) versus after (when it doesn’t, because if you’re still around, you can probably get really good healthcare). Assuming that all post-AGI health risk is erased, then one might divide by around 5 to get the pre-AGI health risk (10 years of risk versus around 50). In that case, the risk is more like a micromort per 7.5 cigarettes or 2.5 liters of wine. So being a regular smoker, or a severe alcoholic, seems like it would cost a few micromorts a day.

What about cryonics?

This one is very speculative. Assuming that AGI is definitely happening in the next 20 years, then my best guess is that the risk of someone born around the year 2000 dying before AGI is around 2%. Conditioning on that happening, my guess that the brain is mostly intact is around 25% (as many causes of death, like motor accidents, might destroy or damage the brain). My guess that utopia happens and its inhabitants succeed in “bringing back” people preserved using cryonics is around 10%. So, the chance that one gets “brought back” is 1 in 2000, which means that not signing up for cryonics costs around 500 micromorts.

The cryonics calculus makes a lot more sense for older people, whose chance of dying before AGI is higher, and whose brain is more likely to be mostly intact conditional on having died.

Conclusion

Surviving until AGI seems pretty worthwhile, and it seems cheap to cut down on major sources of personal risk, at least if you’re relatively young. I hope this post reduces many micromorts and helps people make better decisions. And please let me know if I’ve neglected to mention any major source of micromorts that could be easily avoided. Good luck surviving!

1. Both of these numbers are taken straight from https://micromorts.rip/, but I don’t know what the source is.

2. The source for this one is not Wikipedia but this blog post which doesn’t clearly source the number as far as I can tell.