
Starlight and Sunburst's Misadventures in Creating Artificial Intelligence
Two nerds experiment with AI, but as it advances they find it increasingly difficult to control.
Keeping up with the latest magical research coming out of Canterlot Academy, Sunburst finds something interesting and decides to try it himself. After Starlight sees what it can do, she comes up with ever more ambitious ideas for how to improve it.
They have no idea what they're doing.
Thank you to SparklingTwilight for editing.
Cover composited with HAL 9000 and vectors appropriated from davidsfire and Chrzanek97
Inspired by the work of AI safety researcher/advocate/communicator Robert Miles (featured several times on Computerphile) and by the research paper Concrete Problems in AI Safety by Amodei et al..



7430 words: Estimated 37 minutes to read
3 Chapters:
- Naughts and Crosses 2021-08-04 02:35:26 UTC2262
- How About a Nice Game of Chess? 2021-08-04 02:35:30 UTC2967
- UniVAC 2021-08-04 02:35:35 UTC2201