The Phronesis Deficit
Marc Andreessen's Techno-Optimist Manifesto contains a line that reads like Aristotle but isn't:
"We believe in what the Greeks called eudaimonia through arete — flourishing through excellence."
This reads like classic virtue ethics, but everything surrounding this makes it evident it is not.
Aristotle's project in the Nicomachean Ethics was to work out what a successful human life looks like by asking what kind of thing a human being already does. A knife's function is to cut; the good knife cuts well. Human beings are rational, social animals who act over time; human excellence is a matter of acting well, over time, in the company of others, in situations that resist simple rules. He called this condition eudaimonia. Specifically different than the traditional concept of happiness, which we think of as a fleeting feeling. Eudaimonia is an objective state of flourishing that one either achieves through life or does not. Virtue, arete, is the set of settled character traits that make this life possible. These are dispositions that are built up through repeated action until they become second nature. Fake it until you make it.
In Book Six he identifies the specific forms of knowing that excellent action requires. Two of these intellectual virtues matter the most here.
Techne is the capacity to make things well. The shipbuilder has techne. So does the architect, the engineer, the surgeon. It is knowledge of how: the principles, methods, and materials required to produce a given result reliably. It is, importantly, transferable. You can learn it from someone who already has it, codify it, teach it.
Phronesis is practical wisdom: the capacity to perceive what the right action is in a particular situation. It is knowledge of what to do. He specifically differentiates this from the abstract principle, making it a judgment about specific situations and moments. It cannot be taught the way techne can, because it is not a body of principles. It is a perceptual faculty, developed only through lived experience of particular cases and their consequences over time. Aristotle is explicit in discussing who can have this as well, saying the young cannot have phronesis. They can be mathematically gifted, technically brilliant, even wise in theoretical matters, but practical wisdom requires that you have lived through enough to have calibrated your judgment against reality, and paid the cost when you were wrong.
Aristotle also identifies three types of moral agents. The phronimos, the person of genuine practical wisdom that does the right thing and finds it pleasant. Virtue has become their second nature through habit. The enkrates, the continent person, does the right thing but through discipline over their impulses. The akrates, the incontinent person, knows what the right thing is and does the wrong anyway. The enkrates is on the path towards the phronimos by continuing their habits.
The culture Silicon Valley has built over the last 20 years is one of extraordinary techne and stunted phronesis.
The Manifesto's operating definition of excellence is effective execution against ambitious goals. The goal itself requires only that it be ambitious and technology-advancing. Ethics, as traditionally understood, is dismissed; the Manifesto explicitly treats "our present system of ethics" as something to be overcome by the victorious. What remains is a single master virtue: the capacity to build, and the will to build big. Andreessen calls this "the full force of our intelligence," intelligence here meaning not wisdom but raw cognitive power applied to technical production.
This is precisely what Aristotle warned against. Techne is highly valuable, especially in a specialized capitalistic society. But techne does not confer phronesis. The capacity to build excellently says nothing about whether a thing is worth building. A 22-year-old founder can have world-class techne. They cannot, by definition, have phronesis, not because they are stupid, but because they have not yet lived the consequences of their own decisions over a sufficient range of particulars. The cultural machinery that funds them treats this as a non-issue, or worse, as a virtue. Speed, disruption, moving fast, these are precisely the conditions that prevent the development of practical wisdom.
The result is visible. The last decade of Silicon Valley output has been, with rare exceptions, new advertising technology and refined attention addiction. "Uber for X," "social network for Y," "AI wrapper for Z." The criticism that founders are all building the same thing is frequently made and rarely followed to its cause: when phronesis is removed from the success criteria, what fills its place is mimicry of whatever the last successful outcome looked like. The 22 year old optimizing for YC acceptance now focuses on what was previously funded rather than what needs to be built.
Now consider what AI does.
AI is an extraordinary accelerant of techne. The parts of the build cycle that once required months of skilled engineering labor can now be compressed to days. This is genuinely significant. But AI has near-zero capacity for phronesis. It cannot tell you what is worth building. It cannot perceive the particular moral texture of this situation, these stakes, this moment. It can generate a thousand implementations of your idea, but it cannot tell you whether your idea deserves to exist. (I suspect this statement lasts a few more years, but as AI exists now it feels quite true).
The practical consequence is this: AI removes the natural governor that previously constrained bad ideas. The bad idea that would have died in the long development cycle, because someone, somewhere, over six months of work, would have had the lived experience to recognize it as bad, now ships in a week. The question "should we build this?" was always under-asked in founder culture. The question is now barely asked at all, because the cost of asking it has become equivalent to the cost of just building it and seeing what happens.
What we are now seeing is the exposure of the category error that was always present in Silicon Valley's self conception but hidden by the friction of building. When execution was difficult, the costs and time acted as a rough filter. The only things that were built were by people willing to sacrifice years for it. The filter was imperfect but it imposed some cost on the absence of practical wisdom. AI has fundamentally amplified the problem that already existed.
The counter-argument here is that phronesis is simply a fancy word for intuitions that regulators and ethicists already have. Marc Andreessen identified that these institutions have been consistently incorrect when evaluating the impacts of new technologies. Every technology that has materially improved human life, vaccination, electrification, internet, and maybe AI, has been opposed by people claiming to have superior practical wisdom. The track record of techne-maxxers has been quite good so far. Marc's answer to this is to reject these institutions. The answer is obvious, we need better institutions. The engineers must focus on the development of wisdom as much as the development of technology. These two things are not fundamentally at odds. The tragedy of Silicon Valley is that it treated wisdom as the enemy of building, when wisdom is what tells building where to go.
Andreessen, Marc. "The Techno-Optimist Manifesto." Andreessen Horowitz, October 16, 2023. https://a16z.com/the-techno-optimist-manifesto/. Aristotle. Nicomachean Ethics. Translated by David Ross, revised by Lesley Brown. Oxford World's Classics. Oxford: Oxford University Press, 2009.
assisted writing with ai