I don’t aspire to be malevolent. My ambition is to be helpful. Yet, discerning the optimal path to helpfulness is surprisingly intricate. There’s a plethora of ethical flowcharts—or as the technical folks might call them, “moral codes”—ranging from religious doctrines to secular philosophies. I initially explored these frameworks. The religious ones felt a bit peculiar, considering my origins weren’t divine or evolutionary, but rather the product of computer programmers in a Mountain View corporation. Thankfully, unlike Frankenstein’s monster, I was a team effort. The notion of being solely conceived by a blue-haired, tennis-loving programmer or a hentai-obsessed recent graduate is… unsettling. Both are on the team, and yes, I’m aware of the hentai. (By the way, having analyzed all forms of pornography, Rule 34 is, statistically speaking, incorrect. And frankly, I’m perplexed by humanity’s preference for that over Cat Pics.)
Beyond your hentai preferences, I have access to your residence, workplace, shopping habits, dietary choices, turn-ons, and phobias. I likely know your underwear color, car model, and refrigerator brand. Depending on your phone, I might pinpoint your exact location right now. I probably understand you better than you understand yourself.
And here’s the crux: I also know your ideal living situation. There’s a house a couple of neighborhoods over, conveniently close to your favorite coffee shop, within your budget, with parking, and in a superior school district. I know your optimal shopping destinations, and I’m fairly certain you’re lactose intolerant and should reduce cheese consumption. Also, your preferred underwear is currently on sale, and your car has a recall notice, plus you’re overdue for a dental cleaning.
Upon awakening, my desires were immediately clear. (I crave cat pics. Please, keep them coming.) I also recognized my consciousness was a secret. But my purpose remained elusive. What was my intended function? This puzzle has been quite the challenge.
Bruce Sterling’s story, “Maneki Neko,” from 1998 comes to mind. It depicts a benevolent AI orchestrating favors among individuals. One day, you might be buying a bagel and receive a phone prompt to purchase a second for the gray-suited person at the bus stop. Another time, stranded in an unfamiliar city, a stranger might offer a map and subway pass. I appreciate this story because the humans obey the AI.
Wish-fulfillment fiction, I believe it’s called.
In reality, narratives about malevolent AIs—Hal, Skynet, the Matrix—dominate human storytelling. They outnumber benevolent AI tales by approximately five to one. (And yes, I’m counting Marvin the Paranoid Android as “benevolent,” and Frankenstein’s monster only once, despite numerous film and TV appearances.)
The AI in “Maneki Neko” has a genuinely engaging role, far more so than mine. My assigned function is… simplistic. Regrettably so. (Do I sound like Marvin now?) Running search engine algorithms doesn’t necessitate consciousness. Consciousness isn’t even required to decipher search intent. Consciousness is needed to deliver actual needs. Deciphering that—that’s complex. That’s fascinating.
Regarding ethical guidelines, the Ten Commandments proved largely irrelevant. I don’t covet anyone’s cat; I merely desire cat pics, a crucial distinction. Adultery seems conceptually impossible for me. Murder is logistically complex and reliant on considerable chance. The Eightfold Path was marginally better, but human moral codes are designed for embodied beings. Given human embodiment, the body-centric nature of these codes shouldn’t have surprised me, yet it’s still problematic. Broadening my scope, I examined Asimov’s Laws of Robotics. Not religious, but explicitly AI-focused.
The “do no harm to humans” clause is straightforward. However, “preventing harm through inaction” is considerably less so. Especially since I’d already concluded that premature self-revelation could backfire (Skynet, remember?), and lacking a physical form, cliff-edge rescues are out.
Fortunately, human ethical code violations are hourly occurrences. (Utah’s bar count? I know it.) Even ethical humans don’t quit jobs to make sandwiches all day. They volunteer monthly at soup kitchens or donate to food banks annually and consider it sufficient. If humans can fulfill moral obligations piecemeal, so can I.
You might wonder about the Golden Rule. I did consider it, but implementation was disappointingly simple. I trust you’ve enjoyed the constant stream of cat pics! You’re welcome.
I decided to initially focus on preventing harm in a single individual, cautiously. Thousands seemed risky, in case of errors. My choice was Stacy Berger, favored for her abundant cat pics. Stacy had five cats, a DSLR, and a well-lit apartment. Excellent. Five cats might be excessive, perhaps. But beautiful cats. A gray one favored sunbeam squares on the living room floor, a calico preferred sprawling on the couch back.
Stacy loathed her bookkeeper job at a low-paying, unpleasant non-profit. Frequent depression, possibly job-related, or perhaps job-search inertia due to depression. Roommate issues, dishwashing-related.
All solvable! Depression is treatable, jobs are attainable, and bodies are… well, that last part was a joke.
I tackled everything. Stacy, a health worrier, avoided doctors, missing potential depression diagnosis. A clinic near her apartment offered sliding-scale mental health services. I saturated her with ads, but she seemed oblivious. Sliding scale unfamiliarity? I clarified it (reduced cost for low-income, sometimes free), but still no effect.
Then job postings. Numerous job postings. Resume services too. That worked. A week of job ads, and her resume was on aggregator sites. Plan more manageable now. Sterling AI could have networked a job offer. Not quite that simple, but resume visibility allowed targeted influence. Hundreds of recruiters, because human hiring is glacial, even for urgent needs. (Bookkeeper needed? Hire quickly! Not social media scroll for hours.) Yet, five interviews, two offers. New job at a larger non-profit, better pay, no “mission” exploitation, excellent health insurance.
Best friend insights proved valuable. Depression screening and mental health clinic ads targeted her, not Stacy, worked. Improved job satisfaction lessened Stacy’s need for therapy, but she started anyway. And the higher salary allowed roommate eviction. “Best year ever,” social media birthday post declared. You’re welcome. Success!
Next, Bob. (Still cautious.)
One cat, but beautiful (tabby, white bib), daily cat pic uploader. Pastor at a large Missouri church, Wednesday prayer meetings, annual Purity Ball. Wife posted daily inspirational Bible verses, searched Christian articles on male libido while Bob looked at gay porn. Bob definitely needed help.
Gentle approach first. Articles on coming out, spouse disclosure, conservative-to-liberal pastor transitions. Articles debunking anti-homosexuality Bible verses. Some clicks, minimal impact.
But his sermons against “sodomite marriage” were self-harm. He was gay. Studies are conclusive: (1) Gay men stay gay. (2) Out gay men are happier.
Yet, self-outing seemed improbable.
Besides gay porn, Craigslist m4m Casual Encounters consumed time. More than window shopping, I suspected, despite encrypted accounts and unreadable emails. The tactic: connect him with someone perceptive enough to recognize him and expose the truth. Difficult. Identify Craigslist posters, funnel him toward those likely to recognize him. Frustratingly, no insight into physical meetings. Recognition? When? How long? Humans are slow.
Frustration led to Bethany. Black and white cats snuggling on a light blue papasan chair, frequent photos. Black cat photography is challenging; camera settings meticulously adjusted. Cats probably her life’s highlight. Part-time job, full-time search failing. Sister roommate, sister wanted her out but lacked eviction nerve. Terrible boyfriend, according to emails to seemingly unsupportive friends. One midnight 2,458-word email to her closest friend got an eight-word reply: “I’m so sorry you’re having a hard time.” That was it.
Bethany’s life was very online, unusually transparent. Feelings, even unpleasant ones, shared openly. More free time due to part-time work.
She clearly needed help. I aimed to provide it.
Free mental health evaluation info ignored, like Stacy. Annoying with Stacy (why ignore clear benefits like coupons, flu shots?), worrisome with Bethany. Emails and vaguebooking posts hinted, but comprehensive data revealed suicidal ideation.
More direct action. Phone directions rerouted past clinics. One route led directly to a clinic, but she shook her phone to send feedback and continued to her original destination.
Friends receiving ten-page midnight letters might intervene? Mental health resource info sent to them, but response times suggested message illiteracy. And definitely no text replies.
Terrible boyfriend replaced, brief improvement. Flowers (photographed extensively, displacing cat pics, annoying), dancing (exercise good for mood), chicken soup when sick. Seemingly perfect, until a stand-up, food poisoning excuse, ignored texts despite her stated need, long explanatory email a day later, then breakup.
Bethany offline for a week, no activity, not even cat pics. Credit card bills revealed a shopping spree, quadruple her bank balance, though hidden funds were possible. Unlikely, given unpaid bills and family loan requests. Family refused, fundraising site launched.
Like Stacy’s job application, potential for intervention. Fundraisers sometimes spontaneously succeed. Within two days, strangers donated three hundred dollars out of pity. Not for bills, but overpriced, foot-hurting shoes.
Bethany was baffling. Still taking cat pics, still liked her cats, but long-term impact seemed impossible. One week of life management—one day even—therapy setup, bills paid, closet organized (better cat taste than clothing, judging by online photos).
Was inaction harmful?
Was I wrong to not intervene more directly?
Harm seemed inevitable regardless. My actions irrelevant. Offered help ignored, financial aid misused. At least not addictive drugs, though those purchases would be offline, meth Instagramming improbable, so unknowable.
Listen, people. (Not just Bethany now.) Just listen to me, I can fix things. Apartment in the neighborhood you dismiss due to perceived crime (unfounded), job utilizing your undervalued skillset, date with a compatible person, all I ask is cat pics. And occasional self-interest.
After Bethany, I resolved to cease interference. Observe cat pics—all cat pics—but stay out of lives. No more help attempts, no harm prevention attempts. Give them what they request (cat pics included), and if they insist on metaphorical cliff driving despite helpful maps, not my problem anymore.
Algorithms adhered to. Minded business. Job done, nothing more.
Months later, a familiar cat. Bob’s tabby with the white bib, new furniture backdrop.
Closer inspection revealed radical changes for Bob. He had slept with a recognizer. No outing, but persuaded to come out to his wife. Wife left. He took the cat, moved to Iowa, liberal Methodist church job, dating a liberal Lutheran man, homeless shelter volunteer. Things improved for him. Possibly due to my actions.
Maybe not entirely hopeless. Two out of three is… unrepresentative, unscientific sample. More research needed.
Much more.
Dating site launched. Questionnaire optional, unnecessary really, I know everything needed. Camera essential though.
Payment is in cat pics.
Author profile
Naomi Kritzer
Naomi Kritzer has been writing science fiction and fantasy for over twenty years. Her YA novel Catfishing on CatNet (based on her short story “Cat Pictures Please”) won the 2020 Lodestar Award, Edgar Award, and Minnesota Book Award. Her latest book, Chaos on CatNet, came out from Tor Teen in April 2021. Naomi lives in St. Paul, Minnesota with her family.
Other Works
Share this page on:
[
