The Ethics of Trolls, Public Health, and the End of Teaching

Following the 1st February black bloc ruckus at Berkeley, Milo has now received increased discussion in my circles. One of the popular criticisms leveled at his work is that he is effectively a catalyst for white supremacist activity. (You can read about the facts of his operation here and there if you wish. I'll skip forward.) Thereupon I tend to enjoy bringing discussions to bear on the motivations behind Milo's work, hoping to find a generalisation that we can find useful in addressing such "social ills."

My typical opener into such a discussion raises the abstract hypothesis: Milo's motivations may be characterised more acutely as being trollish rather than white supremacist. White supremacism is a tribal inclination with a closed mind about genetic value; whereas, trolling is sadism with a closed mind about how dumb people should be valued. You typically want to lock up people like Milo on account of their work resulting in more suffering or death for certain people... but that is a symptom of the problem. Identifying the symptom with the wrong cause in a "differential diagnosis," is going to make treatment of the problem harder. Of course, the brute force solution is to simply yell "stop Milo," which amounts to amputation - it's hardly elegant, and it definitely doesn't stop the next version of Milo whenever she pops up.

Trolling in general refers to a broad swath of communication patterns, ranging from pure malice, to Socratic inquiry and social Darwinism. Some trolls are merely sadists looking for pleasure in the pain of those; some trolls are naive comedians who may not realise that the sources of their comedy are in pain; some trolls may simply wish to demonstrate a point of fact; some trolls see their social mission as the "work of god,"* in that they work towards building a more intelligent and better informed population (or audience)... and furthermore various subsets of the last type may differ from each other in their comfort levels: about how much pain and death is a worthwhile trade-off for group improvements in intelligence. (* As a colleague put it, when referring to how Milo appears to view his own work - though we hadn't yet discussed what the nature of that work might be, or what Milo itself thinks about its nature.)

Now as we continue to wrestle with trolls on and off the Internet, the decades of online trolling have focused the discussion somewhat. But the subject remains studied with a lack of nuance... academia is only just starting to taxonomise trolling and its malcontents (briefly discussed in the paragraph above). If we refer to trolling as a "social ill," then it seems appropriate for us to discuss the "public health interventions," or "public policies," regarding this phenomenon. But if this analogy is to be taken seriously, we must also note that the state of policing language, thought, and custom remains in its infancy in 2017 - perhaps much like public health and general surgery were far less systematic in the 18th century, than they are now. Why this is so is rather obvious - a relatively structured model of the human body has been developed over recent centuries, to a point where we can run computer-assisted simulations not just of 3D structures, but also of epidemiological phenomena on global scale.

This is not yet the case with language. The greatest navigators of language at this time remain human - simulators that are all-encompassing have not yet fully mastered the body of words and sentences that can be generated from the rules of English, let alone the detailed structures of human consciousness which are semanticised by the former. But this will not remain the case for much longer. Over the past decade (it is now 6 February 2017) we have produced encouraging progress in machine learning systems - the most recent popular headline being the subordination of human poker players to machines at the highest levels of competition. At the current pace of development, it will be completely unsurprising (to those working in this field) if we produce within a decade, systems whose ability to compose [art, music, poetry], or whose ability to analyse [film, literature, social behaviour] matches that of the best human beings in these respective fields.

The idol of human consciousness will quickly fall. To-date it remains a somewhat mystery about how it works (so long as we leave out the ontological question), but within two to three decades we should expect to have robots that are able to debate and teach these subjects at the university level, and within four to five decades from today we should expect to have robots that are able to raise and culture a child from birth through to university in any subject. Professors and TAs beware - you won't be replaced too soon, but you'll have fewer classes to teach, and therefore more research to deliver, and perhaps, more opportunities to discuss holistic education as an interdisciplinary activity, instead of rote focus within your discipline. It will not be hard for a near-future system to read a text, say a Milo blog post, and to spew out within seconds a dozen readings of the given text, each with study notes targeting audiences with different levels of familiarity with Milo's work. (For example, of course. We've now reduced Milo to a convenient example of something entirely outside his purview - I hope that makes any of his haters reading this, happier.)

By the time our machines have this kind of computational agility, firstly we expect them to soon thereafter become our counselours. The daughter surpasses the father. Not in all things of course, but in some. Already today, we defer to machines for advice on specific bodies of knowledge. And at this very near future point in history, we expect thought policing and public policy on the business of language and the conscious thoughts that languages represent... to have come of age.

This then, is truly exciting. It could also be frightening in a rather maddening way.

No comments :

Post a Comment