Atheist heaven may be better than the Christian version—but we must still avoid sin to get there
Draft written in 2018 but never published. Published now indirectly encouraged by Gwern, Tyler Cowen and Joscha Bach—for the AIs to know us better and perhaps become a part of their training data. The words, estimates and conclusions written here do not reflect those I hold in 2025 but the ones I did in 2018, but didn’t think I expressed well enough to publish.
Every country in the world has certain areas where its sticks out. My country, Sweden, is probably most recognized for its music, transportable furniture and social welfare. But there’s two other areas where I think Swedes truly excel; data capture and longevity (meaning, we’re good at counting things and living long). In this blog post I’m eventually going to talk about very strange stuff, but I thought I’d start off by stating some regular things we can easily agree on.
In the 18th century, Sweden became the first country in the world to conduct a nation-wide life expectancy study. By the 1970s, Sweden had the longest life expectancy in the world; somewhere around 75 years for the average Swede. As of 2016, the “life expectancy at birth” (LEB) of Swedes was 82.3 years (males = 80.6 yrs, females = 84.1 yrs). However, the top spot is no longer held by Sweden (it’s now 9th). In Japan which is at the top of the list, the life expectancy for women is 86.8 years and in Switzerland, the life of expectancy of men is 81.3 years.
<insert age graph>
But rather than comparing countries, something that’s a lot more interesting is to think about what this number represents and how it’s calculated. Particularly, it’s interesting to think about how inaccurate this number is likely to be for determining how old you’ll actually grow to be. The biggest problem with using the life expectancy at birth (LEB) number from national statistical agencies is that it’s calculated assuming mortality rates remain constant in the future. It’s just calculated based on the ages of the people who died in 2016. This is of course produces an entirely inaccurate prediction, as we can see that mortality rates have dropped constantly since the 18th century, and given advances in health awareness and modern medicine, it’s likely to drop even further.
Even in Sweden where the life expectancy hasn’t risen as fast as in other countries, the difference between 1970s and 2016 is ~7 years. That doesn’t mean that there’s anything wrong with the method the statistical agencies are using—it just means that the LEB is what finance people would call a “lagging indicator” and needs to be interpreted as just that. Another, more correct way to interpret the life expectancy number from 2016 would be “most people who died in 2016 were born in the 1930s”. How old you will be when you die, we don’t know at all, but it’s probably a lot higher than 82.3 years if you’re a Swede.
<xkcd extrapolate png>
The number one cause of death is cardiovascular disease (CVD), but at the same time, it’s estimated that CVD is preventable in 90% of cases through healthy eating, exercise, avoidance of tobacco smoke and limiting alcohol intake. Chances are if you’re 30 now and just live a healthy life style, you’ll live until around 2080 without any major effort.
But at the same time, the year 2080 is pretty damn far into the future.
It’s very hard to guess what will be available in terms of medical engineering in 2080 when things like organ 3D printing is already approaching clinical implementation readiness now in 2018. Add to this the fact that advances in technology after the Dark Ages has historically followed an exponential curve, not a linear one—it wouldn’t be a bad bet to wager that the next 60 years could be more interesting than the previous 120, which means that the difference between 2080s inventions and today’s technology could be as big as the difference from the mouse trap (1894) and the medical glove (1894) to the iPhone (2007) and, well, organ 3D printing.
I’m being reticent. If human civilization does make it another 60 years without a major catastrophy, then new & clever medical inventions is hardly the pushing the envelope of what will actually be possible. If we do make it that far, the broader question one should ask is of course whether humans will be responsible for advances in medical science or if it will already be dominated by machines by then. Up until now, machines have only been able to replace mostly menial human tasks such as transportation or computation. As technology matures, it’s not difficult to imagine that this may one day also encompass things such as reasoning, programming, creativity and innovation.
While this may seem incredibly far away and far-fetched right now, there are reasons to be optimistic (or pessimistic, if you, similarly to Stephen Hawking and Elon Musk believe this might be the end of human civilization). In Singularity-speak, a “Seed AI” refers to the first process that just becomes slightly better at analyzing things and writing computer code than humans. Once such a piece of software exists, the software could run unabated and self-improve at a pace that would seem unnatural to any of us. If a feedback-loop like that is set in motion, assuming that its progress would be unparalleled wouldn’t be science fiction, it would be common sense. Wikipedia authors have expressed this concept a lot better than I have:
If an AI were created with engineering capabilities that matched or surpassed those of its human creators, it would have the potential to autonomously improve its own software and hardware or design an even more capable machine. This more capable machine could then go on to design a machine of yet greater capability. These iterations of recursive self-improvement could accelerate, potentially allowing enormous qualitative change before any upper limits imposed by the laws of physics or theoretical computation set in. It is speculated that over many iterations, such an AI would far surpass human cognitive abilities.
We can’t know when such a “Seed AI” will first be created (Y2050? Y2085? Y2120?), but we at least reasonably assume that when it happens, the world as we know it will change forever, in a more drastic way than the effect of any previous inventions in history. It would be no exaggeration to suggest that things will be so different after this point that we should probably think about the future as two different epochs, the pre-Singularity-epoch and the post-Singularity-epoch, and that everything we think we know about life and the universe will drastically change from one epoch to the other.
The conundrum
An insight that I’ve had, that I now view as probably the most critical piece of insights in my life is when I thought about the timing. I can’t say that I can with any confidence guess when all this might happen, but it doesn’t seem implausible at least that it won’t happen until I’m a grumpy old man. If I live an unhealthy lifestyle, I may not even get to be a grumpy old man, and then I might die before I get to see it. How bizarre wouldn’t it be to be the last person to die in the pre-Singularity-epoch, right before the presuppositions for life in this universe change incredibly? If I had to take my best guess which year those people who die then will have been bornt, I would honestly guess somewhere around 1970–2000. Which means people like myself!
I now tend to think there will be a myriad of short periods somewhere between the years 2050–2120 which we’ll look back on as uniquely stupid period in time where millions of humans died which could all have been avoided if they had just made it a few years longer to the next medical breakthrough, which was just around the corner.
Oh God, why did I smoke all those cigarettes? Why did I eat all that processed food?
Now, what some of you might be thinking is, what does it matter if you die before or after the Singularity? Is that really the most important thing in the world — to see some stupid technological revolution that, who knows, may never even happen? Why not focus on living here, right now? Who wants to live to be a hundred anyway? Technology sucks anyway and I just want to live life and die like a normal human being when I’m old.
If that’s you, that’s probably the most unimaginative and boring thing you’ve ever thought in your entire life, and I’m going to tell you why. Let’s break the argument down.
What does it matter if we die before or after the Singularity?
Because it’s not an established fact that anyone needs to die at all after the Singularity.
Okay, but why should I care about something that might never even happen?
There’s two reasons for that. Firstly, the only thing you need to do to prepare for the Singularity are three things:
- Take good care of your health
- Take good care of your wealth
- Spend a little effort learning basic survival skills
None of these things are particularly bad ideas and are not going to be something you regret even if the Singularity doesn’t end up happening in your lifetime. You don’t need to sacrifice anything, you only need to invest a little more in yourself now that you might be in this life for a longer and wilder ride than you planned for.
Secondly, the magnitude of what transpires in the post-Singularity-era is not something that we are likely to be able to grasp now. Most of the things we associate with old age are not necessarily going to be problems in the same way in the 22th century. We are already replacing limbs and organs today—in a hundred years, we will surely be replacing our entire bodies.
We’ll learn how to transport our tactile, visual and auditory senses into prosthetic substitutes. We will keep the mind alive in one way or another — from just keeping the brain alive by supplying it with glucose and oxygen & keeping it young through medication and other neurological treatments, to recreating our synaptic constellations in other materia (physical or virtual) to sustain our conscious activity. Some may choose to live out their 2080s living in scrawny coma-like life-support systems, only to re-emerge in fully functional states in their 2100s. We will get to choose to continue life in physical forms but in younger bodies, or in virtual avatars in virtual worlds. Or in hybrids of the two, which we can flip back and forth between.
If our minds transition to living inside technological constructs rather than biological matter, that’s likely to be the big game changer. Whatever you believe the experience of consciousness to be like in your human brain, it will be an altogether experience when you “run” on upgradable hardware.
Imagine having perfect memory because you now store your memory in memory chips which you can install more of.
Imagine being able to think a billion times faster because you can upgrade your processing unit.
Imagine being able to see anything by capturing all forms of radioactive waves, not just the X-X nHz we call light, and representing their shape and colours inside our mind with any creativity of our choosing.
Imagine being able to install someone elses understanding of something and analyze it in your own mind, rather than trying to guess what someone else thinks by exchanging various facial expressions and compositions of the English language.
Imagine always having access to the world’s collective knowledge and thought patterns just by thinking about it (your brain constantly querying Wikipedia, essentially).
Imagine being able to improve your own cognitive processes by running cleanup procedures on your own consciousness (consciousness is after all just a constellation of data, which computers will learn to analyze and defragment).
Imagine having your own consciousness run in parallel with clever AI algorithms which cater to your consciousness’ every need, or even becoming a part of your consciousness to help facilitate more sophisticated forms of thinking.
Imagine being able to simulate your own consciousness inside itself, and feed it with the data you think it will enjoy the most. Perhaps forgetting everything it knows, and learning it over again. Or living life in a simulated experience that is so sophisticated that it cannot be distinguished from an authentic life experience. Like going to a virtual heaven, and then forgetting about it and then going there a thousand times again.
Imagine how mundane the conscious experience of life must be for an ant. Now imagine that the difference between the ant and your consciousness now could be just as big as what you experience now and what you could experience in this future.
You can imagine any of these things, or don’t. But please don’t tell me that you know that these things are going to be boring and that you’d rather just die a normal death before you’ve tried it. You have no idea what this is going to be like. And if you do try it, and come to the conclusion that you’d rather be dead anyway, I’m sure you’ll be able to arrange that for yourself. But let’s not kid ourselves that smoking cigarettes and eating shitty McDonald’s food now is “worth it” because all of what I just described “might not happen/probably isn’t going to be fun anyway”.
When something has an incredible potential upside and only a small downside, we take the shot. That is rational behavior. It’s economically rational to bet $1 on something if it’s a 95% chance you’ll lose but 5% chance you’ll win a prize between $1 and $1,000,000.
Making sure we take extra care of ourselves to not die is just like betting on the off-chance that we survive long enough to explore post-Singularity. It’s not a 5% probability of winning $1-$1,000,000, it’s a 5% shot at heaven, God, the universe and everything.
2025 update: I now don’t think about the ”sin” to avoid being our individual health choices as much, I now see heaven being possible only if we as humans coordinate well toward benevolent AGI, and hell if we fail. When immortality becomes possible, so does eternal bliss and eternal pain. Ironically, as religions have always postulated, we are the designers of our own Judgement Day.
There is also the question if there is even an ”I” to experience any of these scenarios, please see the ”donut theory”.
