DE EN

Master Thesis

Artificial Fututure

Year

2022

Technology

Artificial Intelligence

Stable Diffusion

GPT-3

Webdevelopment

Goal

How does an AI imagine the future? - Creativity is a very human ability. It is often used as a criterion for being human, especially as a way of distinguishing oneself from other living beings. But will it continue to be reserved only for humans?

Abstract

Artificial Future is an artificial intelligence that is trained to generate its own visions of the future. Users can share their personal visions of the future and work together with the AI. The AI develops its own understanding of the future and the wishes that people associate with it. This knowledge enables it to create and visualize its own visions of the future.

Try it

Try it for yourself @ www.artificialfuture.de

How does an AI envision the future?

At the heart of Artifical Future is a web platform that allows users from all over the world to describe their ideas about the future. It is possible to enter keywords as well as entire sentences or more extensive pieces of writing. The texts are completed and expanded by the AI. At the same time, the system extracts the most important content in the form of keywords and generates an image based on them. In this way, it brings the very personal vision to life.

Create your vision of the future. Powered by AI

Based on usage data, the AI continues to train itself and improve its understanding of a coming world. Finally, individual perceptions feed the AI's ability to create a general, comprehensive vision of the future. It attempts to determine a community consensus by identifying and weighting structures and patterns. The outcome of the AI is completely open and determined only by training by means of the user:in. Thus, it is also impossible to predict whether the vision of the future predicted by the AI will be positive, negative, gloomy, or happy. Likewise, it will constantly evolve, progressively change and adapt.

The following visions were thought up, generated and visualized by the AI completely independently.

GPT-3 &
Stable Diffusion v 1.5

The web application operates through several interlocking subsystems. The GPT-3 language model completes and analyzes the user input. The resulting keywords are visualized by Stable Diffuion v1.5. They are simultaneously used to train the AI model.