This famous philosophical quote was made by Hegel, Georg Wilhelm Friedrich Hegel, and is highly cited and highly confused. We’ll explain shortly but in the meantime let’s first get to the full quote:
Did Hegel say it?
Yes. G.W.F. did in fact say this in his 1920 book The Elements of Philosophy of Right.
Who did he say it to?
The quote appeared in a published work.
When did he say it?
The book came to print in 1820.
What Did Hegel Mean By The “Owl of Minerva”?
Since this Hegel we’re talking about, nothing (neither references nor arguments) is obvious. Therefore, to understand this quote one needs to understand one of the fundamental pieces of this quote: The Owl of Minerva
What is The Owl of Minerva?
The Owl of Minerva, for which there is now an academic journal that bares its name, was a key symbol in Greek Mythology (you may also have heard of it as the Owl of Athena). Hegel is playing off the use of the Owl as a frequent metaphor for wisdom and insight. Further, the Owl was used as a metaphor by none other than Aristotle in his classic philosophical work Metaphysics.
Essentially, depending on the point one is trying to make, the idea of the Night Owl (who are blind during the day) play a key metaphorical role in the history of Philosophy.
What Is Hegel Talking About?
Often, just the following part of the quote is cited and used “owl of Minerva spreads its wings only with the falling of the dusk”, to the exclusion of the full quote (above) or the context (also above). Putting it simply, Hegel is saying that an era can only be evaluated after it has concluded. In other words, Philosophy must wait for the actual actions to take place before it can evaluate what has transpired or opine on it. In general, this is how moral philosophy, including applied ethics, shakes out: we wait for some event to take place and then evaluate the options now that an act can occur. One must be very careful about trying to draw conclusions about their era while still within it.
Another example, a fun fact often asked by tour guides in planetariums, is the answer to the question: what galaxy in the observable universe do we know the least about?
The answer is the Milky Way. Other galaxies we can view in detail with telescopes because we are outside of them peering in. However, the Milky Way is the closest to home but, in many respects, the biggest mystery.
Hegel in the 21st Century
If you’re evaluating this in the 21st century, you’re probably thinking what about moral reasoning prior to the event taking place? Quite a bit of work has been done on moral antecedents which seems necessary for a world that utilizes such epistemic tools for gathering knowledge such as machine learning. To steal another quote from a machine learning specialist, “Machine learning is a philosophy, it’s a way of knowing” that is how computer scientist, and head of machine learning at Google, Peter Norvig begins his crash course class on machine learning. One of the most inspirational things I’ve noticed in my years working in tech is that everything contained in that brief quote is actually true. It’s an epistemology that, when it clicks, it’s unlike anything encountered before. However, it’s also a path to knowledge that requires something of ethics, namely the correct ethical outcome it desires the machine to produce in advance, that ethics may be unable to provide.
The problem is that, without an agreed-upon normative framework, ethics tend to work under the opposite directional flow: we wait for a particular case to occur (Davis, 1992), weigh the relevant information, and then make a call (or propose the options for how different normative approaches might handle a particular case). The relationship between A.I. and ethics can only be rectified by ethical considerations and determinations made prior to the implementation of the predictive algorithm. Ethical conditions that, under the current model of society, are all too often tabled until an ethical violation has been breached (or potentially breached). However, as I’ve outlined, this is diametrically opposed to how A.I. must work to solve problems. Now let’s look at an applied case.