Better than AI

My recent column about my plans to harness AI in my Professional Writing class at the University of the Philippines provoked so many inquiries and reactions that I thought I’d share more of my ruminations, preparatory to a lecture on AI and writing that I’m scheduled to give at a conference in November. Feel free to disagree with and to interpellate me – some statements I’ll be making here might sound contradictory or even outrageous to some good people out there – so I can refine my arguments further.

I’ll begin with the observation that the eminently practical Singaporeans have a strategic plan for promoting AI literacy toward what it calls “AI bilingualism,” i.e., domain expertise plus AI fluency. They don’t think AI should be taught as a separate course but embedded into what people are already doing. Citizens above 25 can avail themselves of free six-month subscriptions to AI apps. Singapore also hosts data centers under very strict conditions.

Can and should we go Singapore’s way? We’re still embroiled in the question over whether to allow a big data center like Pax Silica in – and rightly so. There are many key technical and ethical issues to sort out having to do with resource demands, legal frameworks and people’s rights.

That said, the realist in me expects that AI will be here to stay, whether we host it or not, like it or not, and there have audibly been many of those “nots” lately, in both public forums and social media.

I can understand and respect this resistance. As an education professor studying AI reminds us, skepticism isn’t ignorance. It isn’t just technophobia; adopting AI isn’t as simple as getting used to a VCR, a cellphone or even the internet. Rejecting AI doesn’t mean you’re a Luddite, or you don’t understand algorithms, LLMs and whatever it is they’re gushing about as though the heavens had parted and God’s open palms came out of the clouds showing “A” and “I.” I can see why even tech-savvy writers and artists will refuse to use it in their work on principle.

I myself use Claude in my nonfiction mainly to organize material, but beyond Google, I certainly won’t employ it in my fiction, precisely because I’m doing my best to avoid being predictable, which AI is in a sense all about.

(Am I not horrified by that image of Anthropic’s machine chopping up and digesting books for AI’s training? As an author and antiquarian book collector, of course I am. Will my outrage move the needle? Not by one degree; the harvest will continue, and its output will be so homogenized that we’ll be using it without knowing it. I think the original travesty, if you will, happened when those billions of trees were pulped and fed to the paper mills and printing presses so we could publish our precious books – but of course we don’t think of it that way.)

That doesn’t mean that in the future, writers and artists of unimpeachable integrity can’t or won’t find ways of integrating it organically into their work. It’s scary to contemplate, but AI will mutate to a point of familiarity and trustworthiness that it will be near impossible to deny or to ignore. Taking a page from physics, the controversial but eerily prescient futurist Raymond Kurzweil has anticipated a crucial juncture called “the Singularity,” when human and artificial will merge, likely through nanotechnology, and even boldly gives a date when that will happen: 2045. I’ll be long dead by then, so I’m not even going to worry about its implications.

But I will predict that in the lead-up to that point, at least in publishing, three streams will emerge. At the top tier will be 100 percent human-written books, which will come at a premium though not necessarily sell better (as it is, right now, many if not most such books don’t sell, anyway). In the middle will be hybrid human/AI work – the new and intriguing frontier we have yet to explore, which will be initially met with distrust if not disdain, but will come into its own. At the bottom rung will be 100 percent AI-generated work, especially popular fiction, produced by skilled prompters working for publishers who will be paying them no royalties.

As I’ve often asserted, the notion that “human is always better” simply isn’t true. To begin with, the world is full of brutish, stupid and even murderous people for whom the word “humanity” seems purely a biological description; granted, some such ogres become great artists, but the world is also full of kind, smart and pleasant people who can’t write or draw for the life of them.

In other words, the bar is so low that AI doesn’t have to do very much to surpass human capabilities. It can’t out-write Shakespeare or maybe even the class genius, but it surely can compose a better business letter than most ordinary people – who will be AI’s real base, not academics or creatives wringing their hands over whether and how to incorporate AI into their work. AI mainly empowers the inarticulate; the genuinely good will likely eschew it, or limit its use to the most rudimentary tasks like formatting.

AI cannot be the truth-seeker. It can’t even be the standard against which we rate our students’ or employees’ performance. AI doesn’t work 12-hour shifts or struggle to pay the rent or commute in heavy rain. AI doesn’t cost overhead, leaves, benefits and profit sharing; it’s never absent, and doesn’t talk back, indulge in water-cooler gossip, or comment snarkily on your OOTD (at least not yet). In fact, it doesn’t cost a thing beyond a basic subscription, easily recovered from the savings on paychecks. Sure, it doesn’t have “soul” or that “uniquely human touch” that poets and painters love to wax romantic about – but honestly, did you get those the last time you called Customer Service and demanded to talk to a real live person? How many people truly manifest “soul” and a “uniquely human touch?”

Studies have already shown that AI will consistently outperform people in certain specific tasks. What we need to find is what we can consistently do better than AI. The key question I’d like my students to ask themselves is, “Why should they hire me, and not AI?”

Better than just resignation, my current position is one of critical acceptance: use it where it helps save time and effort with routine processes (but make sure students understand those processes first) and with organizing material following some supervening human design.

As soon as I say that, I know I’m idly hoping that people will resist having AI do their “thinking” for them, and will treat AI as a mere assistant or executor. In many cases, the temptation to have AI do everything will simply be too great. But this is where the most important element of human agency comes in: not even in the design of the project itself, but in the decision if and how to use AI for best results. If you can write better than AI, that might be the best result yet.

*      *      *

Email me at [email protected] and visit my blog at www.penmanila.ph.