Simba
- Berlin, Germany
- January 2021
- Alexander von Humboldt Institute for Internet and Society
The Simba Text Assistant is a web-based tool and browser extension that produces summaries of German-language texts. It is designed to improve users' online reading experience or support users on their language learning journey.
- Open source
- 40/60/0 (% m/f/d)
- Public
Project stage (in ):
Beta/Testing
Implemented by:
Inhouse
Industrial Sectors:
Information and communication, Professional, scientific and technical activities, Education
Usage of AI:
Natural Language Processing
Generation of AI:
Deep Learning (CNN, Transformers, etc.)
Model training:
Supervised Learning, Transfer Learning
Contact
Responsible Person
Freya Hewett
Motivation and values
What in your view defines the public interest and how does your project meet this purpose?
For me the public interest means something that is beneficial for a general public but also specifically citizens that are usually marginalised or face discrimination. Our project aims to 1. provide a usable tool which helps people with lower literacy skills (for whatever reason), 2. provides open source details on a tool which could be used by charities or other companies in contexts that tend to be under-funded (NGOs for example, associations that work with people with learning difficulties)
How did the idea of your project come about?
Internal in our organization
What is the goal of your project in relation to the public interest?
The project is open-source and should be accessible to as many people as possible. The project's goal itself is to make written information more accessible to more people.
Did you follow one or more guidelines for ethical AI, and if yes, which one?
HIIG PIAI :)
Guiding Values
What are your top 5 guiding-values for the project? Top 1
Transparency/ openness
What measures do you use to implement this value? Top 1
Documentation online (GitHub, website)
What are your top 5 guiding-values for the project? Top 2
Justification: why this project, why AI
What measures do you use to implement this value? Top 2
Consulted literature, stakeholders
What are your top 5 guiding-values for the project? Top 3
Technical standards
What measures do you use to implement this value? Top 3
Testing of tool
What are your top 5 guiding-values for the project? Top 4
Collaboration
What measures do you use to implement this value? Top 4
Talked to other people/organisations/users
What are your top 5 guiding-values for the project? Top 5
Usability
What measures do you use to implement this value? Top 5
Related to technical standards, but are generally interested in creating something that should ideally be used by and be useful for people
Design & Safeguards
Which stakeholders were involved in the process of development and implementation of the project?
Developer, Academic researcher
How did you engage relevant stakeholders?
networking (conferences for example)
Did you apply specific methods of participatory design, and if yes, which ones?
no
Have the project results been validated by third parties?
No
Have the design and the results of the project been made transparent to the public?
Yes
In what way have the design and the results of the project been made transparent to the public?
By academic researchers, By the open-source community
Involving the people who will be affected by the project is a necessary part of the project design
1
Which direct channels for feedback exist?
possible to give feedback on our tool, we also have an email address (contact information is visible on website, github, online presence)
How did you verify the accuracy and robustness of your system?
for accuracy: a combination of manual and automatic evaluation. we evaluated in-house, using the most modern metrics from the relevant literature. robustness: small-scale testing (we downloaded the tool and tried it out ourselves)
Which transparency measures do you use to document your use of data and possible biases or limitations of the data sets?
Model cards
What technical and organizational safeguards have you implemented to protect personal data and mitigate possible harms? Choose all that apply
Encryption, Maintaining an up to date privacy policy
Do you take measures in regards to ecological sustainability of your project? Please elaborate.
not really, however we do use fine-tuning methods (like LORA) which are more efficient than full fine-tuning
Does your project rely partly or fully on the use of open data and/or publish results in an open data set? Please elaborate.
we use a partially open dataset (the full dataset is closed, parts of it have been published in a restricted way)