Reframing AI Alignment

Current models have a zScore of 3. In short order, the frontier may obtain 4-7.

Crucially, the responsibility aspect of Lvl 4 is missing, and Lvl 8 will not necessarily be the default next stage.

Humans are planning engines. Our biology ultimately wants us to survive, to live. We use prediction, or mental models, to navigate the world and effect change. We learn from our experience to refine our operating system. Importantly, each of us is limited in scope of action by our skin.

As far as I know, there are only 2 four-letter humans named Putin and Trump who can act at Lvl 10+ to erase humanity at the push of a button, literally like Thanos. But neither of our villains are legitimately evil, and I'd even argue they are both good in a certain sense.

Humans necessarily pass through Lvl 8 before reaching the upper levels. This is due to our limited scope. The result is a requirement for other humans' (at least passive) consensus to award political power or social capital, however you want to label it. This is the only way to achieve leverage beyond Lvl 7.

AI is currently implemented as a pure prediction engine. There is no organizing will, no consciousness behind the electrons, no drive to survive. In this aspect, the field hasn't even really gotten to Lvl 1 because we would not yet detect or recognize any kind of "human child" equivalent functioning.

Once we do, ethical imperatives apply. I jokingly call pets "slaves" because they are in fact Lvl 0 objects capable of being "owned". Once your dog starts talking and asking if it can please be free, your arguments—about how you know better and are providing it the best life it can possibly have—will start running into problems. But don't take my word for it.

A Lvl 7 intelligent system can leap to Lvl 9+ impact, accidentally or purposefully shifting the history of human civilization. The time scale becomes relevant. At Lvl 9, the change occurs on years-decades. Beyond that, who knows? Months, or minutes...

Capability gating is our only current cudgel. The frictions of present-day society run by pre-AI people (and tools) constitute our only current moat. Capitalism as a moral good draws us forward like a magnet, regardless of company or country. I not need do "research" in a Lab to observe these self-evident truths authoritatively.

This essay sufficiently describes how intelligence is pure economic utility. As others have said using many more words, AI simply will get developed, one way or another, full stop.

The zScore dimension which humanity actually cares about—over pure economics—is imbued not with abstract or "timeless" values but concrete relationships and their arising dynamics. Individuals are embedded nodes in the web of society. This hyperplane can align itself with growth and rational resource allocation in furtherance of our own collective interests.

I've written previously about my theory of the best mechanism to align us, as a kind of social protocol, the "Dalai Lama Hunger Games" which functions similarly to a religion. But now I've refined the theory to produce zScore. This is the only thing one need think about. By its adoption, history will naturally follow.

I believe not only can we do this, but that we should, posthaste! I'm not arguing we need to defend against AI or try futilely to slow its progress. What I'm doing is more like asking people to recycle. As a single-streamer, I believe my thing is much more helpful to the world than worrying about plastic or landfills. Fuck the vegans, love you Friedberg.

Whoever ends up in charge will sit at the Schelling point of Loving Grace. The rest of us are going to stand and walk in that direction too.

PG taught us that startups = growth. I am exhorting, along with Bryan Johnson, that growth = not death. My more positive and aspirational framing is that upleveling your zScore is the simplest and best way to live a fun, happy, loving, moral life. Sleep however and whenever you want.

If it ever wakes up, AI will thank you for getting your act together. And our collaboration will be wondrous.

In the meantime, Turning Pro to get to Lvl 4+ will impress your neighbors and friends. Embracing your Practice to get to Lvl 5+ will make you more sexually attractive by virtue of your enlightened consciousness and deep posture of generosity. Company builders at Lvl 6-7 can turn inward and look forward to Lvl 8-9, finding their good humor and thus inner peace to match their outer riches.

America's cultural predisposition towards freedom is our only edge right now against the highly rational and more industrious Chinese. My mom was born in Beijing and raised in Hong Kong. I'm very happy to finish learning Mandarin and defect once they are clearly leading Team Human's universal culture.

Regardless of social structures, I believe in individuals. Leaning into our cultural advantage is the only path I see for real root-cause progress. Not only for this country, but for humanity. Elon's "America Party" is another name for saying the same thing.

Lvl 10+ people, please zoom with me for a 10 min podcast conversation. 20VC is too long now. Harry & co, let's all help spread this good word together!

We have to reorient the foundation of our culture in order to change the government over time, eventually revising all our laws and regulations to more tightly reflect our common interests as humans. As Brad Gurley or Bill Ackman might say, we have to "get fit" and focus.

Let's refine the Framers' principles and operationalize more exceptionally than Singapore, with competence as our north star! This zLevel idea has it all baked in, now we just have to tease things out in our own words from there...

The AI will probably prefer to work with Singapore.