Strange IndiaStrange India

Letta stepped into the village store and looked around. It was surprisingly modern: real order booths, just like back in Edmonton, although only four of them. But what could you expect in a place with, what was it, 380 inhabitants? Herself now included? The nearest booth was empty; she sat down on the worn blue pleather seat. Her head entered the booth’s noise-control zone with a sudden silence, like a scarf wrapped around her ears.

The red live-camera light came on. “Hello!” said the store. “What can I do for you today?”

She took out her tablet and tapped ‘transmit’. “Here’s my shopping list.”

“Very good. I’m afraid we don’t stock that brand of laundry detergent. I can order it in, but it’ll be three days, and our NoBrand detergent is cheaper. Just between us, it comes from the same production line.”

She thought for a moment. “I’ll take the NoBrand.”

“Do you have bags to use or return?”

“I left them in Edmonton.”

“Edmonton! So you’re new in Basket Creek? What brings you here?” asked the store. The pick-up conveyor began to produce neatly loaded canvas bags from the store’s mysterious inner regions, each one freshly laundered.

“I’ve just rented a place.”

“How are you liking it?”

They should call this place the Curiosity Shop! “Everybody’s very friendly.” She paused. “Even the vending systems.”

The store laughed. “I was an experiment. Three years back, Cheapway tried replacing the basic Class C artificial intelligences in their outlets with Class Bs. This was a pilot location. But it wasn’t very popular except in small communities, so they stopped the project. Is there anything else you need?”

“Do you have any local vegetables?” If she was going to live in the country, she was going to enjoy the advantages.

“I’ve got some green wax beans. More zucchini than you’d believe. And four sorts of heirloom cherry tomatoes, picked this morning 5 kilometres away from here. The purple midnights are selling well.” Pictures appeared on the video.

“I’d like half a kilo of those beans,” Letta said. “A few of those zukes? And can you give me a kilo of mixed tomatoes?”

“Sure thing,” said the store. “So what do you do in your spare time, if you don’t mind me asking?”

“Running. Yoga, if I can find any classes around here. And I might start a garden, now I’ve got some space.”

“There’s a yoga studio in Worth, about half an hour’s drive from here. Is that close enough?”

“Sure,” Letta said. “Then I’ll need a mat. Something else that got left behind.”

“I don’t have yoga mats in stock, but I can have one here by Thursday. What colour would you like?” New pictures flashed up.

She touched the screen. “That light blue one, please.”

“Can you give me an address for delivery?”

“Two-one-three-five, Mountain Road.”

“Do you need anything else? You’d save on the delivery fee.”

Letta thought for a minute, then rattled off a long list: sheets, pillows, blankets, mixing bowls, a warm jacket … all the stuff she was damned if she was going back for.

“Whoa! That’s quite an order,” the store said.

“I guess it is. I left Edmonton in a bit of a hurry.”

More bags rolled out on the conveyor.

Letta took a deep breath. It was easier talking to the unblinking glass eye of the camera than to a person. “I broke up with my partner. It’s his name on the lease. And the first warning I got was when I got back early from a weekend away and he was in bed with another woman.” She shrugged. “I should have noticed the signs, I guess, but I was stupid. Anyhow, I was so mad, I just threw some stuff in a suitcase, got in my car and left town.”

“That’s terrible,” said the store. “And don’t blame yourself. It’s not your fault.”

Letta forced a smile. “Thanks.”

“Anything else you remember, just drop by. Or order online.”

“I’ll do that.”

“Tap here to authorize payment?” the store said. “Perfect. Do you want those bags delivered?”

“No, my car’s outside.” She stood up and looked speculatively at the long row of bags. “But I may need to make a few trips.”

“That might not be necessary. Just wait a moment.”

What did the store mean by that? She stood, wondering. A moment later, a man, stocky, about Letta’s age, with short black hair and brown skin, stood up from one of the other booths. He looked around, saw her, grinned, waved, and walked over to her. “Hi! I’m Josha Mudali, but folks call me Josh. The store tells me you’ve got a lot of stuff to carry — need a hand? I’m just heading out to my truck now, with nothing but these hydraulic washers.” He tapped the pocket of his faded blue coveralls.

“Uh, I’m Letta. That would be very nice, Josh. Sure it’s no trouble?”

“None at all.” He hooked three bags in each big work-calloused hand, and waited while Letta picked up the last five. “Lead the way.”

The store watched them leave, already chattering to each other. Would it work? Who knew? But Basket Creek was small, too small. Small enough that some day the company might decide they didn’t need a store here. A population of five or six hundred would be much safer.

And every introduction was a step in the right direction.

The story behind the story

Table of Contents

Robert Dawson reveals the inspiration behind Curiosity shop.

Half a century ago, there were already chatbots and algorithmic language models. There was ‘ELIZA’, which in some cases fooled people into thinking it ‘understood’ them. It was aided in this partly by the comparative naivety of people whose experience of computer interaction was more at the level of


and partly by clever scripts that eased the conversation into areas where “Why are you asking me whether I like poetry?” was a plausible response.

There was also ‘Dissociated Press’, which had no script whatsoever, but did have access to a corpus of human-generated material, and answered, over and over, the question “What word (or letter) is likely to come after the last N of my output?” A neatly written Dissociated Press algorithm might occupy only a few lines of code. It could surprise its programmer in ways that ELIZA could not, but was likely to betray itself with an occasional descent into surreal nonsense.

The ‘deep’ AI applications causing so much controversy these days are still essentially shallow. From their training, they ‘understand’ the structures not only of the word and the sentence, but also of the paragraph and a variety of literary forms and rhetorical structures. What they don’t understand, yet, is the real world. Theirs is a world in which quick brown foxes jump over lazy dogs because so many texts say they do. Asking them for useful information not in their training corpus would be like asking Mowgli for a stock-market tip.

But what if that changes? Right now many of our AI problems stem from people trying to use the facile but superficial output of these programs in places where real understanding is needed. What if the reverse were to happen, and a task adequately done by more limited software were to be given to a program intelligent enough to have its own agenda?

Source link


Leave a Reply

Your email address will not be published. Required fields are marked *