Skip to content
About FNR
Funding
Funding process & policies
News, Media & Events

Luxembourg National Research Fund

PLAITO – Improving the ethical capabilities of software with Artificial Intelligence

 

Artificial intelligence has dramatically transformed our understanding of technology and influenced our interaction with online content and software applications. Whether knowingly or unknowingly, AI has affected peoples’ lives in one way or another. Large language models (LLMs) have been at the forefront of the technological revolution. LLMs are artificial intelligence programs trained on massive data sets that use machine learning to generate human-sounding text and scripts. ChatGPT is a prominent example of an AI-enabled software that has gained enormous popularity over the last few months.

Future forward: Luxembourg’s leap into Artificial Intelligence (AI) is an FNR feature series highlighting Luxembourg’s top A.I. researchers, showcasing findings and results of A.I. research and demonstrating practical applications of A.I. research and its impact on society.

Requirement engineering – Creating a blueprint for ethical AI implementation

Given humans’ increasing reliance on AI and LLMs, questions about ethical concerns have emerged: How inclusive are AI and LLMs? How can we detect and correct biases and stereotypes? How can we regulate and define the ethical boundaries of AI-enabled software? Who is responsible for the ethical implementation?

The field of requirement engineering has addressed some of these issues and concerns. Requirement engineers write, maintain, and validate software requirements to ensure that software complies with the legal and ethical framework.

They maintain and specify the requirements, which in turn, serve as the foundation for regulating digital environments.

You can think of requirement engineers as the gatekeepers between software developers and the public. Just like lawyers set up legal and ethical frameworks in civil societies, requirement engineers specify the requirements under which AI-enabled software operates.

What is requirements engineering?

The term u0022requirements engineeringu0022 was coined in 1964 and refers to the process of defining, documenting, and maintaining requirements in the engineering design process. It is a common role in systems engineering and software engineering.

“Large language models have enormous capabilities. We have to make good use of them for the better of society.”

Dr Sallam Abualhaija is a research scientist at the Center for Security, Reliability, and Trust (SnT) at the University of Luxembourg. Her work centers around Artificial Intelligence, machine learning, and requirements engineering.

She spearheads the PLAITO project, an initiative that uses smart technology, including AI, to integrate complete and inclusive features into digital applications such as ChatGPT. PLAITO aims to build upon existing frameworks and software infrastructures, incorporate less biased insights from generative AI models, and increase the implementation and generation of more inclusive and diverse features.

More specifically, PLAITO aims to help software developers create applications that are more mindful of user needs and more representative in terms of identifying and including software requirements that encompass a diverse set of users.

Large language models are capable of showing us a larger picture of the world in comparison to humans, but we need requirement engineers to ensure that all user needs are respected and stereotypes are combatted right from the start.
Dr Sallam Abualhaija Research scientist in the Software Verification and Validation (SVV) research group at SnT

LLMs and other AI-enabled software applications are mirrors of society. They reflect attitudes, opinions, biases, and stereotypes of humans. The PLAITO project actively combats existing stereotypes, strengthening the existing knowledge database and providing solutions for establishing more complete requirements, thus building more inclusive tools in AI-generated software.

While existing AI databases contain unimaginably large amounts of information, they often have built-in biases and stereotypes. PLAITO aims to enlarge software engineers’ field of vision, meaning it brings potentially overlooked users into their awareness. PLAITO leverages the larger knowledge field of LLMs to give engineers a roadmap that allows them to think about a larger segment of the population and to provide them with practical suggestions to build more effective implantation of inclusive frameworks.

This process could lead not only to software applications being more diverse and inclusive but also improve engineers’ understanding of existing biases and stereotypes.

What is a Large Language Model (LLM)?

A large language model (LLM) is a language model used to achieve general-purpose language generation and understanding. LLMs acquire these abilities by learning statistical relationships from text documents during a computationally intensive self-supervised and semi-supervised training process.

Real-life benefits of PLAITO technology

Dr Sallam Abualhaija uses the concrete example of visually impaired people to demonstrate the potential impact of the PLAITO project: How it could help re-adjust and re-configure existing software, contribute to new software, and provide more efficient and targeted features, such as automated voice chat or more detailed description of video environments.

We need to build platforms that allow cooperation between humans and machines. If we keep humans in the loop, we can take advantage of the enormous potential of AI and LLMs.
Dr Sallam Abualhaija Research scientist in the Software Verification and Validation (SVV) research group at SnT

AI and LLMs are making progress at an extraordinary pace, often surpassing human capabilities in many areas.

Dr. Sallam Abualhaija is committed to building inclusive platforms that enable collaboration between humans and AI-enabled software. Her will help us better understand AI’s capabilities and provide us with concrete applications to reduce stereotyping and increase inclusion and diversity.

Areas that could benefit from the PLAITO technology range from education (i.e., learning apps that provide more tailored content for learning-disabled students), to everyday guidance (i.e., virtual assistance apps such as Siri or Alexa that represent viewpoints of a more diverse population).

The goal of PLAITO is to provide software developers with better digital frameworks, reduce potentially harmful impacts of exclusionary software, and emphasise the opportunities and benefits of AI technology.


Written by John Petit

John Petit is a communication consultant, holding a PhD in the field. His expertise lies in exploring the intersection of technology and society, with a particular focus on Artificial Intelligence (AI) and its impact on our daily lives and broader societal norms. John combines his academic knowledge with practical experience to engage in and facilitate meaningful discussions about the role AI will play in shaping our future.


Dr. Sallam Abualhaija is a research scientist in the Software Verification and Validation (SVV) research group at SnT at the University of Luxembourg. She has been featured in the Research Luxembourg & MEGA campaign “Women & Girls in Science”