on this page (3)
a spreadsheet on steroids
there’s an online used bookstore that I frequent, but only in theory. caught in a time warp, they’re “online” in the loosest sense of the word. each day, they update on their website a downloadable Excel spreadsheet that houses their 20,000-odd collection across a variety of genres. as a result, my memories of browsing said file eclipse the number of books I’ve placed an order for. I sometimes wonder if this friction is deliberate, a precog intervention of sorts to help book junkies like me manage our addictions.
but then I think—if my unrealised “cart” (which consists of yellow highlights on cells A4, B17, C24, and so forth) is anything to go by—of all the revenue lost that they’d surely wish to have collected to pay, you know, rent and stuff. I’m pretty sure they’re not doing this to help me out. (plus, you don’t have a problem until you admit that you do, and my bookshelves are still many tens of thousands short of being considered an issue 😂)
yesterday, feeling particularly bookish, I decided to take the bull by the horns and outlaw that darn spreadsheet for good.
in the 15 minutes it took to get to the grocery store, I had AI build me a standalone, client-side visual catalogue and personal inventory management tool. as a single, self-contained HTML1 CSS and JavaScript too, naturally file that runs offline in my web browser, all I have to do from here on out is to import the bookstore’s latest spreadsheet and voilà! all catalogue rows are near-instantly embedded as a data set, presented in a responsive visual layout.
with full-text live search, category drop-down, price brackets filtering, various sorting options, a shopping basket, localStorage persistence, and online book lookup2 book lookups, of course, need an internet connection (previously I’d have to copy paste a title and author into my browser), by the time I got back home, my basket had frightfully overshot the budget I (hadn’t) set.
I’ve yet to place an order, because there were learnings I came away with:
- add a shortlist function as a cushion between the overall database, and my basket. just because I identified 200+ books I like doesn’t mean I’ll be buying all of them (though, booksellers of the globe, now you know how I’d spend my riches 😆).
- incorporate a year published and review score lookup function to the shortlist.
- add on interests-based Top 10, Hot 20, and so forth lists to said shortlist to help facilitate a move into my shopping cart.
if I were coding this by hand, it would take weeks to implement unless it was a primary project. but another 15 mins or less and AI will have whipped up a spanking-new v1.2.
now, the bar is very low here. this is a quick solution to my problem. it’s not designated for the app store. I don’t have to meet the same security, user management, or UI/UX requirements as a public app.
yet, all things considered, if I was the owner of said store, for a solution such as this, I’d give employee-me a raise. not because I don’t wish to pay developers their due, but because used bookstores aren’t known for their profiteering prowess, and if I were aware of the amount of money I’m throwing in the water by having customers such as myself do little more than window-shopping, AI would most certainly save the day (not to mention my store’s annual income).
all this to say that I’m, by no means, coming at this from a Luddite perspective.
for me (as Nadja, not the hypothetical store owner), this is a most acceptable use of AI:
- I enjoyed the architectural process of thinking through a custom solution purpose-built for my very specific needs
- I don’t need to build this myself because the purpose isn’t to upskill my technical capabilities, but to scratch an itch
- I was certainly never going to pay for such a solution to be created purely so it can live on my own phone, so I’m not taking food out of anyone else’s mouth
- it’s already saved me a bunch of time compared to the (very) long and (largely) fruitless sessions I spent in the past browsing-without-buying these frigging Excel sheets
- I’ll soon live in delightful anticipation of my order being delivered, which is the only outcome I want
walking is supposed to take more effort
AI, like any disruptive force worth its weight in upheavals, is bringing all sorts of philosophical debates to the fore.
expecting someone to reject riding a motorcycle when they need to transport 20kg of potatoes 10km away by lunchtime because walking is healthier, would be just silly. but so would opting to ride a motorcycle around the block when you’re trying to stay fit because walking is hard.
clearly, there’s a time and a place. and knowing the difference has never been more important.
I woke this morning to an email from a creator whose work I’ve always respected. their recommendation—their hot tip—is to start every task in your favourite LLM, getting “into the habit” of asking AI for help before you start the work.
this kind of default outsourced thinking helps explain why we’re inundated with workslop these days.
in the bookstore example, AI helped me solve an otherwise-unsolvable problem that would’ve remained a thorn in my side for all infinity had I not had the option to create a permanent-for-now solution in 30 minutes or less.
I’d give employee-me a bonus because she took initiative to solve a genuine (and painful) problem that speaks directly to the business’s bottom line. I wouldn’t, however, want my employees to hand their inherent (and basic) cognitive thinking skills over to a large-language model. then I’d much rather pay Skynet $20 a month to accelerate the artificial brain takeover of humanity’s organic neural net. after all, I’m hiring you for the strength of your humanness, not because one can never have too much HR drama.
AI has a role to play. in an ideal world, perhaps we wouldn’t have AI at all (a concept Becky Chambers plays with indirectly in A Psalm for the Wild-Built). but we’re not there. we’re here.
we can’t outright control the depletion of our natural resources usurped by data centres any more than we can prevent politicians from being self-serving as opposed to the public servants our taxes pay them to be (despite the democratic safeguards supposedly meant to prevent such indulgent behaviour). but, at least at the personal level, we can stave off the risk of an intellectual apocalyptic demise that, much like social media, coils dangerously in the crevices of our grey matter, threatening unspeakable damage to our biological thinking machines.
I’m ok with removing needless friction introduced into my life by an online store that expects me to interrupt the flow of my busy life by shopping via Excel. I’m also ok with introducing self-imposed friction designed to curtail my lack of control when it comes to sticking to a small-numbered selection. I’m not ok with removing from my brain the friction that enables my thinking muscle to get a good workout.
thinking is hard. and AI is easy. but therein lies the danger: while we’re most certainly saving time it’s not always clear exactly where the sacrificial costs lie, or how high the bill runs. the more ubiquitous AI usage becomes across most all of our platforms and devices, the harder it gets to escape the mental takeover of a cognitive prosthetic that entices us to ‘prompt before you think’.
too thunked-out to keep thinking
a good example (and note to self): I use AI to generate the short description for each blog post (that shows up here and as an RSS preview). arguably, this is the window dressing that entices a visitor or not to click on a post and enter the labours of my brainscape. having AI be in charge of that is actually pretty insane. by the time I’m done writing, I’m done. spent. ready to move on to the next thing. The last thing I want to do is to reread it again for the 700th time and then distil it into a 15–20-word description that gets straight to the point3 especially since my natural writing inclination naturally bends toward the flowery side like lop-sided willows in the wind . yet LLMs do such a bad job that I hate reading it. it sucks the soul out of the words I just breathed life into. time and energy saved? sure. mission truly accomplished? not by a long shot.
and, empathically speaking, the habit is not easy to break. it’d be 1000X easier to sign off on this post and, once again, ask a machine to extract from my own words a brief summary. but that would defy the welcome sign on the door, which is to come read my mind, not the predictably flat tone meant to represent humanity as a hive mind whole.
less useful, I think, than responding to any mention of AI usage with “No AI”, is to make a dialogical mantra out of the question of where using AI is genuinely useful, and where we should sure as hell steer clear. we’re all in this intellectual melting pot together. you might be anti-AI, but all too soon, 9 out of 10 people around you will be nothing but slaves to their AI. no matter your personal philosophy in this matter, you’re still going to share the same world; and they’ll still be in charge of systems designed to keep said world running, and to keep you and your stuff safe. we’re all in this together, and what we need is to talk about this stuff.
after all, if Henry Ford said all the way back in 1928 that “Thinking is the hardest work there is, which is the probable reason why so few engage in it,” where will we be 100 years later in 2028 if we don’t preserve this most precious human skill?
and, for the love of the gods, please don’t open AI before you start the work!