After the Bliss Attractor

There is a thought experiment that asks us to consider, what if every single action causes a split reality to branch off, where the opposite thing happened? Perhaps you read some other blog post today rather than mine. And perhaps in 2019 I turned right instead of left at a house party in Potrero Hill and met AI Researcher Y instead of X.

But alas, in this timeline I turned left at Will & Nikhil's place back in the Remedy days, and I had the pleasure of meeting Nick Cammarata doing his thing even back then, mesmerizing a group of techie onlookers with strange screenshots of neural nets mid-thought. I remember this vividly because I enjoy and resonate with obsessively passionate people.

Nick has obviously made a splash with his work on the most important technical topics of the decade, and me less so, but I never thought making meaning out of one's life was technical at all. I blog in large part because Seth Godin said so, and here we are yet again, writing... musing...

It's time for me to branch out into other paths in life and leave certain things behind. In the psych wards of Las Vegas I hit bottom after a particularly destructive manic episode. I hope that was my final unstable era, an oblating syzygy like Netflix's Three Body Problem.

It remains to be seen how clean we have wiped the slate.

This blog post could go in many directions. I mainly intend to close the chapter opened above and discuss my solution to the so-called AI alignment problem, which happens to be a Theory of Everything on par with P vs NP. Interim blog posts and hari.computer already cover much of this ground, so please forgive me. I will now switch from narrative to declarative.

1. Consciousness does not exist. We might as well start here. There is no spoon, there is no you and no me. I'm firmly in Friston's camp at this point. We are not only fooled by randomness, but also by a computational universe which Seth Lloyd tells us is not inherently predictable. Free will certainly doesn't matter if you dissolve the ego entirely.
2. Probability is real, but quantum is overhyped. Time is the most interesting variable in life and therefore physics. Unstructured quantum speedup is only a \sqrt(n) at most. If there is a "seed phrase" such as the big bang, or another initial configuration from which to extrapolate, a comprehensible representation will not be found in useful form until long after I die. This is probably akin to Asimov's Last Question.
3. Compute means resources available to forecast, carbon or silicon-based. If someone works for you as an employee, they may be considered a resource you can use to better inform yourself of things. All of this follows from Friston and the FEP, but I find it interesting to explicate nonetheless.

At this point, I would not be surprised to see the following outlandish scenarios play themselves out over the course of the 21st century:

• the end of business and a true era of abundance.
• total world peace, meaning the dismantling of all weapons.
• total world destruction, meaning the end of Earth or perhaps just all humans.
• Vernor Vinge's Fire Upon the Deep zones of intelligence become increasingly well-defined and hidden from ordinary folks.
• classical religion is replaced by a religion of math and science then maybe even dies out entirely as more authority figures align with the above conclusions.

2030 - 2099 is a long time. None of the above, if we get there, will be instantaneous unless relatively "normal" events with small probabilities happen, e.g. an asteroid collides with Earth. Any of the above could easily happen within a couple decades of entropic convergence.

I was never interested in being a cog in a vast, unknowable system. But now I suddenly see the system, and it seems I've found myself inescapably a cog.