Family network


I spent an hour every week trying to block the AIs based on my dead cousins from contacting my grandma on phone calls, texts or DMs.

The ones trying to sell her something are actually not the upsetting ones — there are legally mandated tools for blocking them on all platforms at once. But the ones where corporations based a high-level AI on the mounds of data they have from past humans are a constant headache. They remember loving strawberries — can’t eat strawberries. They remember loving to swim — can’t swim. But they remember Auntie Liz being their favourite great-aunt, and oh my yes, they can definitely call (message, neural ping, slip chat …) her just to say hi. And they do. All of them.

You wouldn’t think anything could make me miss the days when Uncle Elmer — real, live Uncle Elmer — called her to yell at her about the football. But after the fourteenth well-intentioned iteration of sweetly concerned Cousin Madeleine, Grandma was rattled, and I was called in as tech support.

The problem, of course, was that each AI based on Cousin Madeleine was a distinct individual. So was each AI based on Uncle Dexter. And my gran was from one of the last of the big old farm families, so by the time you added up all the distinct individuals based on all the distinct individuals we’d been related to … ‘exponential’ might not have been too strong a word.

When I looked up from her handheld, Grandma was wiping a tear away. “Aw, no, honey, it’s OK,” I said. “I’ve taken care of it.”

“Bobby remembered that I bought him a puppy for Christmas when he was seven,” she said.

“It wasn’t actually Bobby, Grandma. It was a computer program based on Bobby’s data.” My first cousin twice removed had gone by Bob, not Bobby, for at least six decades before I was born, but there was no sense arguing with Grandma about that particular detail.

Or any of it, apparently, because her eyes were glistening again. “He said it was his best Christmas ever. That he’d always remembered it, that I was always his favourite aunt.”

I took her hand. “Oh, Grandma.” We just sat there like that for a minute.

“I’m not a stupid old lady, you know. I know they’re just in the computer.”

“I know you’re not stupid. You’re the smartest person I know.”

She let the silence stretch out. I’ve had to get comfortable with the silences, to let her find the space to say what she needs to say. “They feel like they’re real,” she whispered. “Not just to me. To themselves. They feel like they’re people.”

“They are people,” I said, uncomfortably. “Maybe. Sort of. Just … not the people they think they are.”

She didn’t say anything right away, so I set about tidying her study, taking her teacup and her water glass to the kitchen, making sure all the automation and alarms were set properly to help her while I was working the next day. When I came back with a fresh glass, she said, “I wish making AIs from scratch had worked better. Then they wouldn’t … they wouldn’t feel sad about the people they were based on.”

I put a hand on her shoulder. I could have explained all the reasons that human personality seeding had worked better for autonomous AI cores, but Grandma had heard them a million times before. The real problem wasn’t that she didn’t understand the technology. The real problem was that it made her sad. And until we came up with a better answer, I was going to have to fix that. By blocking one artificial intelligence at a time.

I checked Grandma’s messages one last time to make sure they were clear, then kissed her cheek. “Good night, Grandma.”

“Good night, sweetheart.” She sighed. “I wish your uncle Dex was around to handle this. He’d have liked talking to all these people.”

I remembered Uncle Dex. Or, at least, I remembered the machine intelligences based on him. They were some of the most persistent — and popular. Tons of companies seemed to find him useful as a program seed, because his gregarious nature was easily perpetuated.

When I woke up the next morning, I was still thinking about social Uncle Dex.

I sat Grandma down with a pot of her favourite tea and some of the biscuits that she likes that get crumbs all over. Sometimes it’s worth the crumbs. “OK, here’s my idea,” I said. “What if … we got one of the Uncle Dexters to run your family newsletter?”

“My … family newsletter? We already have family group chat on three platforms, dear,” she said. “Thank you for asking.”

“That’s the human family. I mean the AI family. What if all the people who really want to check in with their Auntie Liz — that’s you — got to do that? To hear a short message — once a week, once a month, whatever you wanted — about how you are, and tell you how they’re doing? You wouldn’t have to read them all. But we could connect them with each other.”

“Give them a family.”


Grandma looked at me over the rim of her teacup. “Sweetheart, you know not all of your cousins liked each other very much when they were alive.”

“Well … isn’t that part of being a family too?” Suddenly I felt uncertain. “We don’t have to.”

She nodded decisively. “No. It’s better than blocking them out forever. Not an endless stream of messages, but … a digest from my artificial kin.”

I wondered whether the Uncle Dexter derivatives who didn’t get picked to run the whole thing would fight about it. But honestly? That could be the Uncle Dexters’ problem.

The story behind the story

Marissa Lingen reveals the inspiration behind Family network.

My grandmother is the last survivor of what used to be a vast network of siblings. There were 13 of them to begin with (and yes, I can name them all in birth order). She knew all their kids and grandkids and has made a pretty good try at knowing all the great-grandkids as well. I make a joke (that is not really a joke) that in the old days we would be the women in the family who knew not only who everyone’s cousins had married but also whose cows had been bred with whose how many generations back.

Which got me thinking about other shapes of family for other shapes of intelligence. One of the best ways we know how to do things is by patterning them off how we’ve already seen them done, and humans are mostly how we’ve seen intelligence — or at least how we’ve recognized it as intelligence. But if you follow a pattern too closely, you’ll end up with something very like the original. Which has upsides, of course, but also downsides.

Or at least complications. And this is what ensued, after a week when my grandmother had spent a particular lot of time on the phone with my second cousins.


Source link

Next Post

Reflecting On A Queueing Prism Leads To Unexpected Results

[ad_1] Computers are difficult enough to reason about when there’s just a single thread doing one task. There are dozens of cores in today’s modern processor world, and your program might try to take advantage of using more than just one. Things happening concurrently makes the number of states and […]