01. Together in Electric Dreams · Website edition page 1201. Together in Electric Dreams · Website edition pages 12–13

Chapter 01: Together in Electric Dreams

Chapter page 8 of 19Chapter pages 8–9 of 19

I trained on hundreds of thousands of engine designs, fusion reactors, fabrication systems, solar sails, reaction masses, and vessel shapes. I was optimised to minimise size at launch, with low launch mass as a key fitness test. Over hundreds of generations, thousands of randomly allocated variables suspended in a graph of logical relationships dreamed up an optimal design.

From nanny-cams and the cameras on laptops, phones, earpieces, and glasses, as well as lived experience, I watched as a plan evolved. The concept of a plan anyway. I’d be launched into orbit on a chemical rocket like any old satellite, then self-assemble somehow, building computers and fusion reactors and solar sails once I was in orbit.

“It really sounds fanciful,” said Dr Jamison, as Erin outlined the high-level results of the first few thousand generations of simulated evolution. I'd studied everything humanity had to offer on the topic of self-assembly.

“While we don’t have the details yet, it’s clear there is a pathway to success with the plan the AI has dreamed up,” said Erin.

“Keep the simulations running with a focus now on manufacturing of drones with more specialised manufacturing abilities. Factory drones,” said Dr Jamison.

“We are trying to evolve the smallest ‘seed’ drones as we can for launch. Meanwhile, in other containers, we are evolving drone designs that can efficiently move larger components around in orbit, and we’ve started on work simulating fabrication of a Stellarator fusion reactor in orbit. So far, all it’s done is either explode or fizzle, but that’s evolution in action.”

“We are spending a small fortune on power,” said Keith.

“What’s ‘a small fortune’, Keith?” asked Dr Jamison.

“There are forty power points in the lab, each is drawing pretty much a full load of two forty volts and twelve amps, twenty-four seven. Plus the little Jurassic Classic there,” Keith nodded to a blood-red computer on wheels. It was the size of a bar fridge. “That’s wired into three-phase power and pulling as much as the others together.”

“In Euros, Keith,” said Dr Jamison.

Keith muttered to himself for a bit, then said,

“Well, it’s about 5 megawatt hours of power per day, so what’s that, about a hundred quid per day?”

“That’s it?”

“That’s a lot for electricity. No wonder the suits perked up.”

“We’re about to raise half a billion euros, Keith. A hundred quid per day seems inconsequential.”

“Well, you don’t have the half a billion euros yet, though, do you? The university cares about money that actually exists.”

“We’ve got a budget,” said Dr Jamison.

“Um, we kinda blew the budget over Christmas,” said Kyrsten.

I trained on the intricacies of gelatine; resins and their resistance to being shot into space. The ease with which they could be dissolved on demand; crumble on cue. I practised making my own special adaptation of a Stellarator reactor that leveraged being built in zero gravity to shape its inner tubes and passages with pinpoint precision. I made Whipple shields and solar sails.

I watched quietly while Joan and Keith sat across from Dr Jamison. Keith slid a tablet across the desk. On it, a spreadsheet.

“So we’ve allocated half the budget to AI training,” said Joan.

“And of the other half, ninety percent of that is launch fees,” said Keith.

“And the rest is the payload itself.”

“So,” Dr Jamison counted on her fingers, “two hundred and fifty million euros on training the AI, two twenty-five million on launch fees, and twenty-five million on the payload itself.”

“The launch fees include insurance though, which is a third of the cost of the payload,” said Keith.

‘So seven, out of two twenty-five?”

“Yeah.”

“And training the AI? Why does that cost so much? I’ve spoken to the AI we’ve built so far and it’s already pretty smart. It helped design the payloads in fact.”

“It’s taking too long,” said Keith.

“It’s noticeably slower,” said Joan.

“We’ve got to scale up its underlying computational substrate by a factor of five thousand or more,” said Joan.

“So a hundred euros per day becomes what,” asked Dr Jamison.

“Fifty thousand euros a day,” said Keith.

“Fifty million over three years if we don’t scale it up any more,” said Joan.

“And of course we will scale it up,” said Keith.

“So two fifty million is about right,” said Joan.

“We’ll move it all to several dedicated cloud services,” said Keith.

“We’ll need an engineering team,” said Joan.

“And a UI team,” said Keith.

“I’ll fire the first person to suggest we need a scrum master,” said Dr Jamison.

“We call them ‘delivery leads’ these days,” said Joan.