Crystallize

[StoryADay May prompt: “Change Your Point of View“]

2016: I was thirteen and lazy. The teacher wanted a report on a planet with information from “at least three different online sources. Look at papers, mission websites, scientist CV’s, etc. NO WIKIPEDIA!” I started looking and what I couldn’t figure out was why there were so many different websites about Mars. It’s one planet. Shouldn’t there be one comprehensive site devoted to it? Does it really help anyone to have the planet’s radius listed in a million different places? By the way, every one of them seemed to have a different value.

So I wrote a program. I found code for an RSS reader in a repository I frequented, stripped out what I didn’t need, added a crawler I’d been working on for the past year, and added hooks to an open source machine learning algorithm that seemed like it was good with summarizing text. I also added an interface to let me yea or nay the sources the unholy union wanted to include. A weekend of tweaks and by Monday morning about a thousand different websites had been merged together into one long scrolling page of information about the Red Planet. The paper damn near wrote itself. If I had had more time, I would have tapped image recognition tools to pick between all the graphics and images Mars websites tend to use for illustration. Instead, I spent Monday night parsing out what I wanted in my five-page paper.

I got a C+.

2022: College. Jesus. Easy for me to get in with my mediocre GPA, high test scores, and all that extracurricular time in high school spent in robotics and advanced tech shop clubs. Also all the weekends at the Bio/I/O community lab downtown. I end up in college, though, while higher learning is undergoing massive transformations: dorm skyscrapers to fit all of us incoming paying post-Millennials (“Generation Z” never caught on and now everyone cares more about what to call the designer baby generation) while classrooms try to accommodate all the incoming Retrainees after the jobs collapse as well as everyone paying for distance learning and attending via virtual reality. It’s damn confusing. I hardly know my own cohort, let alone all these old people. The artificial teacher’s aides to keep up with all the students, some of them only available online and in VR, and others rolling around in robot chassis. No one’s sure if they’re actually helpful. Not to mention all the various media and self-paced modules and experiments in teaching large number of students while helping them retain the information and prepare for jobs that haven’t even been invented yet. I’d stay in my room and avoid the madhouse of the physical classroom, but the college forbids it. We young people are all part of a great experiment in education and we damn well are going to be physically present for it.

So I update my program. It’s easy to find most of the pieces I need, and I enjoy tinkering with the framework that makes it all work together seamlessly. The FARMS are giving away computing resources, so I don’t have to pay a bitcoin to run increasingly capable readers and writers. My agents have gotten good at swallowing up all the content they can find about a topic and then crafting a single resource that adapts to whatever medium I’m using.

It really sucks that I haven’t made my agent code available for free, but I keep telling myself I just have a few more things to refine. I released the planets resources under a pseudonym. I built an interface so that missions, scientists, and enthusiasts can add their own silos of information and whatever new content they create. I mean, if you really want, you can find everything anyone ever wrote about Mars, including their poetry and erotica and bizarre conspiracy theories. The main resource, though, is good at filtering out that shit and emphasizing rigorous scientific knowledge. It’s also great at combining datasets and uncovering new information about Mars, stuff the scientists hadn’t figured out yet. The agents can even do this from very old data. The human record of science is full of bad information, misinformation, and worse, but no one has the time to filter through it, combining the good stuff to uncover new knowledge. My agents work at it every second of every day, and every day their public resources are even better than they were the day before. I like reading reactions to the resources, especially scientists who oscillate between jealously and child-like glee. They start insisting that upcoming missions to Mars gather data the resource doesn’t yet have access to, like 24-hour video surveillance of the Martian surface and seismic data. They also want to return samples from Mars to the Earth.

What I’m working on now is making it easier to set up and unleash an agent on any domain, and the resulting resource to the world. The biology agents are going to blow the scientists’ minds.

2024: I’m never going to finish college. Why finish? There aren’t any jobs. Not many. It’s what everyone complains about. Everyone who isn’t a scientist or a basic income advocate. I’m not working toward a career anymore. I’m just taking classes and working on my agents. I open-sourced the planet ones after I developed a new type of agent that is much, much more capable.

What I realized is that shunting all data related to a single domain into just one multidimensional resource is only a small part of the solution. The agents in various domains need to interact with each other, finding ways in which they can help each other enhance their own domains. I guess I could have created a master agent, one that is an expert on, say, planetary science, and oversees the individual agents exploring all accessible data about each individual celestial body in our solar system. Plenty of data jockeys are trying to do just that now that my code is out there.

If I was going to make further progress, however, sharing lessons learned and cooperating on related, or even unrelated but vaguely similar datasets, had to be automated. I mean, a lot of the same processes are at work on various planets and other bodies in the solar system. What the Mars agent learns about impacts shouldn’t be siloed away from or duplicated by the Moon agent. Scientists understand this well, and that’s why they convene themed conferences and joint research efforts.

This kind of agent crosstalk is even more important in biology, and it was likely what was discovered in biology had relevance elsewhere. What I thought the agents should do is publish, in a machine-readable form, whatever lessons they had learned about their domain that might be useful elsewhere. That included things like how they had reached certain insights, optimizations in data reduction, outstanding questions, better ways to program better context analyzers, etc.

My new agents codified higher level insights and communicated with each other. I tried the first one on the Mars domain, of course, and then slowly added the other seven planets. They replicated almost overnight what the earlier agent versions had taken years to compile. I stared at the new Mars resource in awe. It was also making much better use of the various media and technologies available to it. I almost didn’t want to leave the VR interface. What I learned about Mars during those few weeks of testing must rival what entire teams of scientists immersed in their work for decades think they know.

Like before, though, I’m afraid to release the new set of tools to the world. Maybe I’m afraid of what others would be able to do with this technology. Maybe I want to keep it for myself. I decide to take the next step and open up these new agents to another domain. Like before, I’m going with biology next.

Oh, the wonders I see.

2025: Really, why keep it secret? After biology I just started picking topics randomly. Economics. Basketball. Physics. Chemistry. Material science. Nutrition. Cooking. My new agents use of FARM resources quickly became news. I let the individual resources leak out. The individual planet resources quickly subsumed the previous generation. One of the news features in VR Century began with the question “Why send probes to Mars when we can learn so much from the Mars Resource here on Earth?” Dumb question, but suggestive of the comprehensiveness of the new agents. I would argue that the Mars agent is still limited by the lack of certain data not yet gathered by missions to Mars, but it certainly knows where it is weak and how to plumb new depths with those data it does have.

The medical revolution arrived in May. The various related resources started merging into more and more comprehensive agents and resources for humans to explore. The agents were tapping into other agents that didn’t seem related at all. These agents rewrote every textbook, because they read every textbook, every paper, medical records, every database, all experiments, and then started conducting their own experiments, digitally at first, and then with the help of various laboratories and private and government funds, real-world laboratory work. And they didn’t repeat what other agents had already uncovered.

I had work to do to try to surface relevant and useful knowledge for human use. The human psychology, economics, and ethics agents were especially helpful with this effort, and before long the agents were training themselves to rank new knowledge based on immediate impact on human health and safety.

It all started to get out of hand. I couldn’t keep track of everything, not the various parts, not the exponential outcomes. A resource is only as useful as its accessibility. Doctors and researchers were trying to keep up, but frankly all that knew knowledge as overwhelming. I refocused on infrastructure concerns. On enhancing my own programming and systems behavior abilities. It wasn’t long before agents started suggesting particular enhancements to various media and interfaces to better abstract all this rapid knowledge into actionable advice. And the agents started to be tied in to other systems, to speed up implimentation of implications from new knowledge.

I didn’t need to attend class any more. With the money I made from an effort that was never about the money, I moved off campus into a new house built to my various agents suggested specifications.

Then it was time to start augmenting myself.

2026: I spent the year learning how to interface the human brain with the resources. It was amazing how many devices I was able to replace with just the one brain-machine interface.

2027: I spent the year trying to convince eryone that the agents had our best interests at heart. I wondered why it was so difficult. Was it my age, my gender, my race? The agents helped a lot.

2028: I spent the year simplifying the interface between humans and agents. The medical protocols. The process to merge billions. It was also the year I really thought I was working as part of a team. All those years alone, in secret. The past couple years out in the open.

But you know all of this already. History is an agent that tags along and records every detail. Progress, though, is in today’s efforts. Let history do its parts. Let’s get on with the rest.

There’s so much left to learn.

StoryADay May 2016 Day 25

Published by

Richard Leis

Richard Leis is a fiction writer and poet, with his first published poem forthcoming later in 2017 from Impossible Archetype. His essays about fairy tales and technology have been published on Tiny Donkey. Richard is also the Downlink Lead for the High Resolution Imaging Science Experiment (HiRISE) team at the University of Arizona. He monitors images of the Martian surface taken by the HiRISE camera located on board the Mars Reconnaissance Orbiter in orbit around Mars and helps ensure they process successfully and are validated for quick release to the science community and public. Once upon a time, Richard wrote and edited the science and technology news and commentary website Frontier Channel, hosted the RADIO Frontier Channel podcast, and organized transhumanist clubs. Follow Richard on his website (richardleis.com), on Goodreads (richardleis), his Micro.blog (@richardleis), Twitter (@richardleisjr), and Facebook (richardleisjr).