this page is in progress
Reflection¶
This week we have heard a lot of theory on extended intelligences from Pao and Daniel and thanks to their effort to present in an easy to understand way, we have gained a basic understanding of neural networks, machine learning models, pipelines and databases. This enables us to communicate with developers and understand the difference and interplay between different subcategories of AI.
Exhibition¶
I found 4 exhibits at the exhibition at BCCC particularly interesting.
Waterfall of meaning¶
An artificial intelligence about the meaning of individual words. which was also an inspiration for our group project. i think it is particularly interesting here that the ai places the words in a space that is not immediately obvious to us in the first thought when you examine the meaning of a word. it is a neutral way of looking at the meaning of a word without thinking from a biased perspective.
Brainbow¶
An ai that can help to illuminate and trace connections between neurons. This AI illustrates that we can use Ai’s to work more precisely, faster and easier, creating new possibilities.
Slocum G3 Glider¶
Ai combined with soft robotics to perform tasks underwater or in space that humans cannot do. These concepts show the potential autonomy of Ai’s. they can collect, process, learn, store and communicate data in a way we can understand. Of course, the robots have been pre-programmed in a certain way to perform a task that we cannot perform with a human, but they can also adapt to the current environment and make decisions on their own to accomplish the mission. but how well can they deal with unknown situations. can they get creative and find their own solutions spontaneously?
Robots Modulars¶
Group Project¶
In our group intervention we tried to develop a text bias detector. We looked at and tested different databases and pipelines. One pipeline was able to calculate the probability that the entire text was biased. the second pipeline that we ended up using was able to highlight individual words in the input text. This application could help us to write texts ourselves or as a browser extension to analyze texts. However, the database with which the model was tainted was not sufficient. for an application like this, it is particularly important to use a database that covers all perspectives and is maintained by different locations and people. A possible next step would also be to categorize biases and mark them differently.