The idea

We believe that it's possible to improve the natural language message transmission from anywhere.

Using any device with Internet you will be able to send a message to Mars. 

This message will first be transformed into its minimal version using AI and it will be sent to the local orbital satellites. Through the x-band segment of the electromagnetic spectrum they will transmit the message to the orbiters on the red planet, which will relay the message to the Mars ground station.

Finally the AI on Mars, trained in the same way as its terrestrial partner, will convert the message back to its original form.

We implemented a two prone approach:

1.- Infrastructure: having 3 geostationary satellites, 2 in solar orbit (with the same orbit and period as the earth) and 3 in Areostationary orbit (Mars).

This satellite constellation provides:

• Redundant network routing.
• Allows connection even in conjunction periods (Sun between earth and Mars)
• Data transmission at all time (doesn’t rely on satellite coverage availability)
• Scalable to other planets in the solar system

2.- An AI that learns from natural language and creates an evolving map of most common words/phrases to a 4 hex char words. For this event, we created a simple algorithm that read books available from the project Gutemberg and extracts the most used words (more than 100 occurrences)

This demo was able to read 10 books in a fraction of a second, extract the most common used words and add them to a base of 3.000 words that already are recognized as the most used in the english language. After the training, the encoder (called Mars Message Encoder) was able to create a dictionary of words and a dictionary of most common phrases. With this two packages, we encode the test message.

Another section of the demo, shows what happen if there is a new word that the encoder doesn't recognize. It lets the user decide if he let the AI to "learn" this new word. On an advance version of this AI, it should decide if it's worth it to add it to it's dictionary (it will depend on the occurrences between messages and the availability of words) or to directly replace the word to one that already is in the library with the same meaning.

The Future - Using advance AI technologies as GPT-3 (openAI)

AI could predict text based on intentions on short messages and write the full text. Think about an AI that can write an article, a book or a message.

Mars could have an evolving code library aid by an AI that from small text inputs could write code to be used to re-create websites tagged as ”Built for Mars”

All this ideas have the same concept in common. Instead of sending everything, send only a small set of rules or intents and let the AI do the hard work.