I Asked a Neural Network to Help Me Write

I’ve been scrambling to finish the draft of Nahara the last few weeks, which hasn’t left much time or brainpower left for trenchant blog commentary. But it has left time to giggle madly at neural networks.

Meet Transformer, which completes texts one word at a time based on what it learns by reading the Internet. It’s a bit like predictive text, only much, much more well-read (well…it’s read more, anyway).

Being a neural network, however, it’s remarkably bad at some things. In particular, lists.

So of course I asked it to make me some lists. My prompts are in bold.

Birthstones

1. January: Garnet
2. February: Amethyst
3. March:  Mauve
4. April: Pearl
5. May: Ruby
6. June: Topaz
7. July: Citrine (if you have it)
8. July 31 – August 4
9. September: Garnet
10. October:  Amethyst
11. November: Pearl
12. December: Ruby
13. January: Topaz
14. February: Garnet
15. March: Citrine
16. April:  Amethyst
17. March 31 – January 9 (2 days after patch – patch is available, but won’t update.)

Nancy Drew Mysteries

1. The Secret of the Old Clock
2. The Hidden Staircase
3. The Bungalow Mystery
4.  The Secret of the Big Bottom Beds
5. The Mysterious Black Door in the Clock Tower
6. The Disappearance of a Woman at the Clock Tower
7. The Clock Shop Maniac Who Sold His Wedding Ring to Jack Sparrow
8. The Clock Tower, Where the Clockmaker Has Died
9. The Clock Factory
10. The Clockmaker Who Never Came Back

Every Disney Princess

1. Snow White
2. Cinderella
3. Aurora
4. Ariel
5. Belle
6. Donald
7. Tiana
8. Sleeping Beauty
9. Anna or Olaf
10. Sleeping Beauty

US Cabinet Positions

1. State
2. Defense
3. Treasury
4. Energy
5.  Other

So, that’s a complete list of potential threats to the federal government.