Why you should embrace generative AI, and why you should resist it.
The boardroom soap opera of the past week seemed an appropriate punctuation to a year in which generative artificial intelligence went from esoteric CompSci topic to something we all needed to be concerned about. Like a handful of pebbles thrown into a pond, the launch of ChatGPT, barely a year ago, created ripples which continue to expand and overlap. “Knowledge workers,” who for years looked on complacently as digital automation put checkout clerks and assembly line workers and travel agents out of a job, suddenly confronted a future in which they too could be replaced by machines. CFOs salivated over OPEX savings that would come from swapping out messy, inefficient humans with computers. Luddites and other contrarians split semantic hairs, chortling over ChatGPT’s mistakes, and arguing that these models were no closer to “general intelligence” than your smartphone’s auto-fill function. Money poured in. The rest of us kept ourselves entertained asking ChatGPT to write poems and break up with our boyfriends.
I come neither to bury GenAI nor to praise it. Like any technology, GenAI is neither
inherently evil nor inherently good. Scoffing
at it or ignoring it won't make it go away. Deifying it, despite OpenAI’s lofty mission statement, will not bring
about the betterment of humanity.
Granted, there is a lot to worry about. I won’t even touch the hypothesis that a paperclip-maximizing
AI will wind up annihilating humanity. The doomers may be right in the end, but humans
seem to be perfectly capable of coming up with ways to annihilate themselves
without the aid of a large language model. I do worry about the stuff that GenAI models make up as they respond to
a prompt. The term “hallucinations” -- evoking dreams and psychedelia -- is deceptively benign; “lies” is more
appropriate. In 2024, citizens of many
countries including the US will elect their next leaders amidst a swamp of artificial
words and images indistinguishable from real ones, and I think we’re in for
some seriously ugly scenes.
And yet, retreating to our pencil-and-paper sanctuaries and
refusing to engage isn't a good answer. Your job will (or already
does) involve interacting with AI at some level. GenAI models will make your
life easier, especially by automating certain repetitive tasks. Working alongside a GenAI, if you do so skillfully,
will liberate you from a whole lot of drudgery – think writing meeting minutes,
summarizing long documents, cleaning out emails, etc. Rather than doing your job for you, GenAI can
actually enhance the work that you’re doing. To paraphrase medical futurist Bertalan Meskรณ:
professionals will not be replaced by GenAI, but professionals who know how to use
GenAI will replace those who don’t.
Amid all the chatter, I think it’s beneficial to consider a
question that I don’t hear very often when generative AI is discussed: What
do humans lose when they stop writing? What happens to them when they no longer devote time and energy to the
practice of composition, when they delegate that activity to a machine?
Content generation is among the lowest-hanging fruit for companies
looking to improve their P&Ls using GenAI. Without a doubt, ChatGPT can churn out, adapt, and optimize mountains of
text a lot more efficiently than you or I can. The marginal cost of producing another unit of content approaches zero. And a funny thing starts to happen. Words, and the ideas they signify, become
reduced to units of economic output: manufactured, ingested, excreted. The relationship between the source of the content
and its consumer becomes purely transactional. More than that: the “reader” becomes nothing more than a vessel through which content
passes as supply and demand curves seek equilibrium.
I get a similar feeling sometimes when I think about money. My salary, direct-deposited into my bank
account while I sleep, effortlessly sluices through me on its way to my creditors,
thanks to the “auto pay” function I’ve set up. I have forfeited my agency, like we’ve forfeited many things, for the
sake of convenience. We get out of bed
each morning to keep feeding the machine.
Yet I still see glimmers of hope. A few months ago, I was asked to write a one-page memo summarizing a strategy presentation my team and I had put together. This could have been accomplished in a few seconds by uploading our 50-or-so slides into a GenAI model. However, the presentation was confidential, so a public model like ChatGPT wasn’t an option, and we hadn’t yet stood up an in-house model. So I spent two or three hours drafting the memo myself. And in the act of writing it – putting one sentence after another, searching for the right words and flow, crafting a few paragraphs to convey a set of ideas – something interesting happened. I became able to articulate our strategy more clearly. I found that I actually understood the story we were trying to tell much more deeply than I had beforehand.
If you've ever kept a journal, perhaps this sounds familiar. Our appreciation for experiences, situations, people, ourselves, you name it, becomes much
stronger, fuller, realer, when we try to write about them. Writing helps us make sense of our world,
which seems even more important at a time when the world is full of a lot of stuff that seems
senseless. As Flannery O’Connor said,
“I write because I don’t know what I think until I read what I say.”
Writing prose, as opposed to just consuming it, also forces
us to reflect at least fleetingly on the human being who’s eventually going to read
it. We consider our words, not only for
their accuracy, but for the impression we wish them to leave on another person. Writing is a solitary and often lonely
activity, but, perhaps paradoxically, it fosters empathy. Will this speak to my audience? What is their state of mind when they sit
down to read it? How will it “land?” All part of a delicate but critical maneuver which requires that the writer place herself inside the head of her
reader.
When humans stop writing, I would argue, they become less
human. Or at least they become lesser
humans. They get mentally and spiritually
flabby. They become Matrix-like vessels for forces that do not have their best
interests in mind. They think about other people less. They succumb to powers of this world that rebel against God.
So, yeah: understand what GenAI is and what it isn’t. Get
comfortable with ChatGPT. Learn to work with GenAI tools, just like you’ve
learned to use a word processor. But at
the same time make sure you don’t surrender the opportunity to create prose of
your own. Take the time and make the
effort to work with language, to put words together, to create meaning. At the very least, it’s good for your health. At best, as George
Saunders puts it, “purifying one’s prose style is a form of spiritual
dedication; working with language is a beautiful and noble way to spend one’s
life.”
