Postdoc in Sustainable Generative Artificial Intelligence for Media

Tasks

The five-year research project Frugalai examines the effects of using generative artificial intelligence (Genai) to create media content. Current Genai models, which are used to generate music, images, text and other media, are based on huge data sets and require extensive calculation resources. Training and large-scale use of Genai models contributes to significant energy consumption, resource consumption and increased carbon dioxide emissions, which exacerbates the environmental impact. The project aims to develop and promote an innovative and sustainable alternative to today's large, commercial genai system. The goal is to reduce the environmental impact of Genai technology and at the same time promote positive social effects by prioritizing small, ethically collected data sets and local models and calculation resources, while users are encouraged to sustainable behavior.

The research and development within the project can be placed in the field of human-computer interaction and AI (Human-Centered AI). Frugalai has the following goals:

• Appreciate the environmental impact from current Genai systems based on large amounts of data (Big Data).
• Develop a set of tools for resource-efficient ("Frugal") genai open source, based on small and ethically collected data sets.
• Mapping the effects on social sustainability when using resource-efficient genai.
• Comparing the environmental impact from these tools with Genai tools based on large data sets.
• Suggest and apply evaluation measures to take account of resource use.
• Examine the consequences for other Genai applications (eg image generation).

The postdoctor, which has a background within HCI and machine learning, will contribute to all goals over a full-time period. The work is done in an interdisciplinary team consisting of engineers, music scientists and designers, and includes the following tasks:

• Contribute to the development of an interactive environment for webcasts (a prototype is "Stochastic Pirate Radio", KSPR) by designing and maintaining the project software, develop workflows to create media programs and for automated monitoring of offensive content, and SUPPLY
• Research around and developing principles for resource efficient AI in order to reduce model sizes, data sets and calculation requirements using the latest technology in machine learning.
• Develop a measure for sustainability within generative AI (environmental, economic and socio-cultural sustainability) that includes several goals, based on C. Douwe's Thesis * On The Environmental Impact of Deep Generative Models for Audio * (Sorbonne University, 2023).
• Contribute to organize and implement workshops along with team members.

For more information, contact Bob L. T. Sturm (bobs@kth.se).

Qualifications

Requirements

• Deposed PhD or foreign degree deemed to correspond to a doctoral degree. This authorization requirement must be met at the latest at the time when the employment decision is made.
• The project's working language is English, so you should master English well.
• We would like to have a background in HCI and machine learning
• You have good cooperation skills

A merit

• At the end of the date of application for a maximum of three years since the doctoral degree or a foreign degree that is deemed to correspond to the doctoral degree.
• Awareness of diversity and equal treatment issues with special focus on gender equality
• It is merit if you can work independently.

We will add great emphasis on personal qualities.

Become part of kth

KTH forms the future through education, research and innovation. As a leading international technical university, we play an active role in driving and participating in the transition to a sustainable society. Here you are offered the opportunity to grow and develop in a creative and dynamic workplace with good working conditions and benefits. Gender equality, diversity and equal conditions are a quality issue and an obvious part of KTH's values as a university and state authority.

Read more about our benefits and how it is to work and develop at KTH.

Union representatives

Contact information to union representatives.

Application

You apply via KTH's recruitment system. You as applicants have the main responsibility for your application complete when submitted.

The application should contain:

• CV with information on relevant professional experience and competence.
• Copies of diplomas and grades from previous university studies.
• Personal letter where you describe how your background matches the tasks in this service as well as your interest in the overall project. No more than two sides long.

The application must be received by KTH no later than the last application date at midnight, CET / CEST (Central European Time / Central European Summer Time).

If the employment

The employment applies until further notice, however at the longest two years

An employment that postdoctoral is a time-limited merit employment with a main focus intended as a first career step after defense.

Other

For information on the processing of personal data in connection with recruitment.

It may occur that an employment of KTH is located in a safety class. If this is the case for this particular employment, a security review of applicants is made in accordance with the security law (2018: 585) after consent. In these cases, a prerequisite for employment is that applicants will be approved after security testing.

We abandon us direct contact with staffing and recruitment companies and salesman of job ads.