‘The Memory of Tomatoes,’ a short story from an alternate future

We overestimate what we can do in a day, and underestimate what we can accomplish in a decade. Welcome to Alternate Data Realities, from Simply Secure and Consumer Reports with support from the Mozilla Foundation. These science-fiction stories, all set in the US in 2030, reflect a reality in which “we the people” benefit from our own data. Following each story, a response essay from a policy expert offers actionable recommendations on what’s needed for this alternate reality to come to life.

The house wasn’t big enough for Merida and her mother Isabelle.

It had two bedrooms and one bathroom, and Merida longed for an office. When she worked from home in her bedroom, the exhaustion that lived always inside of her—that persistent knowledge of her mother’s worsening condition—lured her back into the blankets.

The backyard had enough space for one raised bed, upon which Isabelle’s window looked out. Whenever Merida tried to grab a moment of peace, her mother rushed out the moment she touched shovel to soil, reminding her of what they’d both lost. But Isabelle had raised Merida and her three sisters in the house, and Merida feared that moving would rid Isabelle of her final, clinging memories.

Merida received a voice message in her earring piece from her virtual assistant, Jenny: the tomato plants needed fertilizing. Jenny’s OS worked across all of her devices, from her earring pieces, watch, and contacts to her home automation hardware and the nostalgic cell phone she used. The fertilizing treatment arrived at her door an hour after Jenny’s alert. When she had downloaded Jenny, she set up automations for the most important home goods, groceries, and medications.

“Mer, your cortisol levels are low,” Jenny said in her ear as she gardened. “Can you make a few management decisions?”

Merida had become accustomed to talking to the AI as though it were a friend. “Lay it on me.”

“A new Alzheimer’s drug is beginning trials,” Jenny said. “They are asking you to donate some of your mother’s medical data.”

“Which data?” Merida asked.

Jenny displayed a list on Merida’s watch, which she tapped up into her contacts and read through. Before Isabelle’s memories had gone, she had fed her own AI with her thoughts and decision-making processes through a series of diary entries, tests, and surveys. Then Isabelle assigned Merida as a filter for her data decisions; there were too many instances of companies preying on the elderly with overreaching requests. Now, Merida had the option of forwarding decisions to her mother’s neural link. Unlike Merida, who preferred devices that she could remove from her body, Isabelle preferred the implant. Merida nodded affirmatively, and Jenny quickly interpreted her head movements and body language as confirmation to send the question to Isabelle. 

“Your mother requested developmental updates on a medical device,” Jenny said. “That information is now available.” Merida sent the info to her mother without asking for clarification; she liked to let her mother keep some things to herself.

“Now we have a bid on your medical data from a company advertising to seniors with memory problems,” Jenny said. “They’re offering a one-time fee of $25 to gather your and your mother’s medical-specific data.”

“What are they selling?” Merida asked, then shook her head. “No, wait. I don’t care.” She declined.

“Now we have—”

Merida pulled her hands out of the dirt. “Sort the rest of my messages and don’t show them to me until tomorrow.” She wiped the dirt on her pants and opened up her oversized phone to more easily scan the list of tasks her supervisors at her marketing firm had forwarded to her. The first was a meeting with their vintage clothes client. Before the widespread adoption of virtual assistants, she might have needed to call a sister in to care for Isabelle, but Jenny gathered data from her sisters and scheduled Merida’s time away to flow with her sisters’ free hours. When no family was available, the hours were offered to a network of approved care providers to bid on the paid work. As she wiped away sweat, her doorbell rang with her sister’s alert. Merida nodded to let her sister inside. As Merida returned her gardening tools to the shed, her mother stepped sleepily into the yard.

“Jasmine?” she said. 

“No, mom. It’s Mer. Jasmine’s in the living room.”

“Merida?” she said. “Can’t be. Merida’s my baby girl.”

Merida loaded her mother’s routine and scanned it; her mother was low on Vitamin D, and a little sun would help, though she needed to take her vitamins. Her mother’s pill case tracked her dozens of medications and vitamins; Merida’s care app told her that Isabelle had taken none today, so Merida would need to bring them to her later. “Have a seat, Mom. I’ll have Jenny load your nostalgia playlist.” The doctors had advised Isabelle to listen to as much music from her youth as possible.

 “Jenny said it’s supposed to rain,” Isabelle said.

“That’s tonight.”

Merida’s mom backed into the house. Merida tapped her mother’s medical schedule and current readings over to her sister’s phone. They’d chosen to keep all of Isabelle’s medical data on their network, for safety. Jasmine was good at helping her mother to remember, better than Merida had ever been.

Merida hugged her older sister. Then, she slid into the oven a pan of enchiladas her youngest sister had ordered. Even though the youngest couldn’t be there in-person, she taught remote music lessons to a local farmer through the neighborhood network; in exchange, the farmer’s husband cooked multiple meals for Merida’s family each week.

         She closed the oven. It would turn off and alert her mother and Jasmine through a green light in the ceiling when the enchiladas were the exact right temperature so that Isabelle wouldn’t burn her mouth: a number of degrees gleaned by Jenny from her mother’s involuntary noises of approval and stored with all the numbers that made up their lives within the backdrop of their days. 

Merida stood outside the clothing store, peering into the window at an outfit her mother might have worn in her 20’s: a bright pair of parachute pants and a spandex shirt.

“Turn offers back on,” she said. All at once, Merida’s current project flashed to life, an overlay on the window display. The personalized offer that she designed projected across the window: “We combine 80’s style with eco-friendly fabrics, sourced locally. Totally tubular!”

Merida laughed; her team had added the slang, and she liked thinking of her mother saying those words. She walked inside for the meeting with a genuine smile on her face.

When Merida left her meeting, an alert sounded in her earring. Jenny chimed in. “Your friend Kellan sent out a request for a walking companion in the park a block from here. Would you like to join him? The daffodils are blooming, and there are exactly four other people there right now.” Merida smiled. She liked that Jenny knew her preference for uncrowded spaces, and that heat mapping technology made it possible to quantify crowds without identifying them and risking privacy violations. “You could check off your daily activity at the same time.”

She agreed to meet her friend then set off toward the park, happy for the chance to clear her head.

Isabelle stroked the parachute pants Merida brought home. “My mother gave me these,” Isabelle said from her recliner. “Is she here?”

“No, Mom,” Merida said. “She’s not.”

Merida gazed into her bathroom mirror. She often felt like she was forgetting something, and she worried that her forgetting was a sign of things to come. As her worrying sped her pulse, Jenny pushed a breathing exercise to the mirror: a bubble that expanded and contracted as Merida breathed and glowed green when she had calmed. 

“Your blood sugar’s low,” Jenny said as Merida’s watch displayed the care instructions issued by her doctor for such an occasion; they were monitoring her closely for pre-diabetes, but the doctor suspected the issue could be corrected with a close eye, a revised diet, and increased exercise, a treatment plan based on the outcomes of other patients with similar profiles. She thought back on the day and realized that she’d gone four hours without a snack.

Merida ate according to her doctor’s instructions. Jenny scanned the food and drink to log it in Merida’s food tracker. Twenty minutes later, the blood sugar alert in the top right of her contacts faded—and she heard a ding that signaled a call from her new friend. She nodded to accept.

“Forget to eat again?” Ken said, his voice coming from her earring.

“How’d you know?” Merida laughed.

“I’ve been there.” He paused. “But also, I followed you on your meal tracker today.”

“Did you? I’ve been so busy I didn’t even check those settings when I downloaded it. I hope I haven’t been eating too much junk!”

“My Jenny knows that I like all the social options,” Ken said. “She’s always asking me to sign up for this site or that.”

“You like to know that much about your friends, do you?”

Merida had met Ken through a group of caregivers she found by accepting an invitation to share her own data—cortisol levels, exercise amounts, diet specifics, and all the other factors that made her tick—with a research community developing products to make caregivers’ lives easier. In addition to the server where they worked together with researchers on co-design and usability studies, the community invited her to a forum where fellow caregivers shared advice and data to streamline decisions they were forced to make for themselves and their loved ones. For privacy, the community members exclusively used avatars. Ken had shared with her a gamification and seed sharing app for gardening, and after they realized their shared love of plants, they traded actual identities off-forum. 

“I was thinking,” Ken said. “I’m supposed to take the bullet train down to your neck of the woods next weekend. Would you like to meet?”

Merida’s heart sped up as she added the invitation into her calendar, where Jenny would schedule a time that worked for Ken, Merida, and Jasmine.

“I’d love to meet.” The decision was the easiest one of the day. 

Merida snipped dead leaves off of her tomatoes. Maybe, she thought, one of the greatest joys in life was pruning, ridding yourself of the rot, the old endless list of to-dos that tried to take up space in your brain. She remembered how it felt to keep a planner, to endlessly plug meetings into her calendar. She snipped another dead leaf.

“I taught you how to grow tomatoes,” her mother said from the doorway.

“You did.” Merida looked up. Her mom wore the parachute pants.

“You brought me these pants today,” her mom said.

“You remember?” 

Her mom knelt beside her in the grass. “I don’t know if that’s the right word.”

“What do you mean?”

“Before I lost it all, I recorded my favorite memories and stored them with a company researching neural stimulation through the link.” Her mom tapped her head. “That’s what I said in the message I left myself.”

“So… what? You’re being fed memories?”

Merida’s mom nodded. “I saw you with the tomatoes from out my window, and I saw the memory, too. The time I helped you plant tomatoes. It feels different than a memory. But it feels kind of nice at the same time.”

Merida struggled to stay composed as her mother reached out and ran her hand along a stalk.

“I taught you well.”

“You did,” Merida said. “Thank you.”

As the rain drizzled, neither Merida nor her mom moved from the dirt. Instead, they let the weather wash over them.

Merida settled down for the night. She thought about memory, how fallible it was for everybody. She might pull up an old video on her phone and be surprised at the details. She might forget tomorrow that she’d agreed to meet Ken. She lived with devices doing the work of memory for her; and was it any different the way her mother had chosen? As with so much, she didn’t know the answer, but she looked forward to considering the question.

---

In response to ‘The Memory of Tomatoes’

By Richard Whitt, president of GLIA Foundation.

A hopeful future of Jenny VAs

In some ways, the central character in “The Memory of Tomatoes” is Jenny, the virtual assistant (VA). Merida has come to rely on Jenny for a vast array of tasks, including filtering all her data-related decisions, running various smart devices, and providing Merida’s business services via an online bidding platform. Isabelle has even uploaded her previous thoughts and memories to her own AI via a “neuralink,” and has delegated to her daughter her data access rights — in this instance, to her “smart” pill case.

One can assume that such trust is well-placed — that Jenny is not busily divulging to random marketers and nefarious third parties the sensitive information she is continually gathering and acting upon. After all, Merida keeps Isabelle’s medical data on a personal network, rather than in the global cloud, and her local community employs IoT devices on crowds in a privacy-supportive fashion. 

On the other hand, Ken admits to having followed Merida via her meal tracker (a bit creepy), because she had neglected to opt out of those open settings. The story also refers to a presumably lawful marketplace for selling one’s medical data. So this world of the near future seems an interesting blend of guarded private spaces, and more intrusive connected technologies.

Paving the way: AI tech, plus human governance

Jenny seems to serve Merida as a virtual repository of trust and support. Nonetheless, a fully trustworthy AI environment is not the one we have today. Over the coming ten years, numerous market, social, and public policy changes will be necessary to open up new space for the alternative ways of combining human data and computation that are exhibited in the story.

For example, commercial interests are exploring fomenting markets for truly Personal AIs (PAIs). As with Jenny, these specific kinds of virtual assistants would act on behalf of their human keeper, and not the corporations who created them. Recent technology and market trendlines demonstrate tremendous promise for the coming development of these PAIs.

But human-centric tech alone is not sufficient. These AIs would need to be provided and supported by trustworthy human institutions. Such entities — whether as for-profit companies or non-profits — would act as our authorized agents, actively mediating between our own personal interests and the rest of the Web. Ideally, these new intermediaries would treat each of us with respect as actual patrons and clients, and not as mere Web “users.” 

A number of governance models are possible. Under one scenario, individualized digital fiduciaries and collective data trusts would serve us, operating under longstanding fiduciary-type principles. These would include explicit duties of care (do no harm, and act prudently on my behalf) and loyalty (have no conflicts of obligations, and promote my best interests).

Seeding the Public Policy Realm

Often public policy conversations about the Web resort to various ways of holding incumbent players more accountable for their actions. This can extend to seeking an express right to privacy, for example, and related protections of one’s personal data. These types of accountability reforms are important and welcome. 

In the world of “Tomatoes,” the main characters possess a more formidable kind of power — the ability to bolster their individual autonomy and agency. They enjoy opportunities to create and control access to a vast array of their data and computational resources, and in turn share that access with trusted people and entities. In order to promote that new ethos of human autonomy/agency, policymakers can adopt a set of digital rights which Web companies must instantiate in their public offerings . 

As the story exemplifies, these rights could include:

• Systems interoperability (Jenny works across all Merida’s devices, and links with other human and environmental devices as well);
• Data portability (Merida shares medical data with a trusted research community);
• Virtual delegation (Merida delegates certain responsibilities to her personal AI, which in turn is authorized to act on her behalf vis-a-vis third parties);
• AI intermediation (Jenny functions according to what can be seen as the “Four I’s” of interop, interface, interrogation, and instruction); and
• Digital preservation (Merida’s desire to feed Jenny with her thoughts and memories presumes that entities must preserve people’s digital objects).

Policymakers can utilize other tools to create appropriate incentives for entities to adopt business models and governance structures based on fiduciary law principles. These incentives can include (1) “safe harbors” from data protection compliance mandates, (2) government procurement requirements, and (3) taxation benefits.