Safety and Bias in Generative AI 

Description

The Data Science Institute (DSI) at the University of Technology Sydney (UTS) is a world leader in data science and ethical artificial intelligence and has a strong focus on Industry engagement. The UTS Data Science Institute DSI boasts over 200 PhD students and 100 academic staff with award winning researchers in AI 

In collaboration with Leonardo.Ai we are advertising an industry PhD program that will tackle research questions at the bleeding edge of GenAI research with an aim to drive forward both foundational research in the field through traditional publication outputs, as well as supercharging the Leonardo.Ai platform. Leonardo.Ai is a world leader in generative AI and is building one of the leading AI laboratories in foundational AI research. Leonardo.Ai has also recently been acquired by Canva and now hosts over 200 million users through the combined platforms. Both Leonardo.Ai and Canva have strong principles in: being a good human being foremost, maximizing impact, moving at speed and owning our goals and winning together.   

Project Aims

This PhD project aims to advance the field of safe and ethical generative AI by developing robust classification and detection methods specifically for AI-generated media. Current safety classifiers are primarily trained on real-world images and video, which often fail to perform effectively on AI-generated content. Addressing this gap, the research will focus on designing and refining safety classification methods for multi-modal outputs in generative AI. The student will therefore push the current safety methodologies into the regime of precision machine learning, such that edge cases are exceptionally rare on large scale multi-modal AI. 

Thesis topics will explore a variety of approaches, including classification systems for generated outputs, analysis of model hidden states, and other metrics for detecting undesirable content. The research may also involve post-training techniques to refine and remove problematic concepts from AI-generated outputs. We aim to open source this work and present the research findings at major AI conferences globally. 

The successful PhD candidate will have access to leading edge compute facilities (e.g. H100s), a team of world class research leaders across both UTS, Leonardo and Canva, cutting edge tools to conduct research, and mentoring and guidance to complete a best-in-class PhD research program. 

Eligibility

We are looking for the best of the best! You will be adequately qualified to enter the UTS PhD scheme with Undergraduate / Postgraduate Qualifications in e.g. Computer Science, Data Science, AI, Physics, Mathematics or equivalent.   

Candidate Profile

• A strong background in programming, mathematics, statistics or related areas. 

• Knowledge and interest in Diffusion, GANs, LLMs and generative AI in general as demonstrated by independent project work in GitHub. 

• Proficiency in modern deep learning frameworks such as Pytorch or Jax.  

• Ideally, we are looking for candidates with previous experience in top-tier Tech companies. We, however, welcome applications from underrepresented groups and diverse backgrounds.

• Proficiency in English (both Verbal and Written). 

Location 

This PhD will be hosted at the University of Technology Sydney and the student will spend significant time at the Leonardo.Ai headquarters, which is located on the waterfront in North Sydney (20 mins train ride from UTS).  

Available Scholarships

Full Domestic Scholarship: Open to Australian citizens who will receive a full scholarship of $37k per year stipend + $10k per year top-up for 3.5 years. 

Full International Scholarship: Open to all international applicants with $37k per year stipend, $45k per year international fees and $30k per year top-up for 3.5 years. Student visas will be provided by UTS.