Women in AI: Emilia Gómez at the EU started her AI career with music

To give AI academics and others their well-deserved – and overdue – time in the spotlight, TechCrunch is launching an interview series focused on remarkable women who have contributed to the AI ​​revolution. We’ll be publishing articles throughout the year as the AI ​​boom continues, highlighting key work that often remains overlooked. Read more profiles here.

Emilia Gómez is a senior researcher at the Joint Research Center of the European Commission and scientific coordinator of AI Watch, the EC initiative to monitor the progress, adoption and impact of AI in Europe. His team contributes scientific and technical knowledge to EC AI policies, including the recently proposed AI law.

Gómez’s research is grounded in the field of computational music, where she contributes to the understanding of how humans describe music and the methods by which it is modeled digitally. Starting from the musical field, Gómez studies the impact of AI on human behavior, particularly its effects on jobs, decisions, and children’s cognitive and socio-emotional development.

Questions and answers

In short, how did you get started in AI? What attracted you to the field?

I began my research in AI, particularly machine learning, as a developer of algorithms for the automatic description of musical audio signals in terms of melody, tonality, similarity, style or emotion, which are exploited in different applications ranging from music platforms to education. . I started researching how to design new machine learning approaches addressing different computational tasks in the music domain, as well as the relevance of the data pipeline, including the creation and annotation of datasets. What I loved about machine learning at the time was its modeling capabilities and the shift from knowledge-driven to data-driven algorithm design – e.g. instead of designing descriptors based on our knowledge of acoustics and music, we were now using our know-how to design datasets, architectures, and training and evaluation procedures.

Through my experience as a machine learning researcher and seeing my algorithms “in action” in different domains, from music platforms to symphonic music concerts, I realized the enormous impact these algorithms have on people (e.g. example listeners, musicians) and I oriented my research. towards evaluation of AI rather than development, particularly on studying the impact of AI on human behavior and how to evaluate systems in terms of aspects such as fairness, human oversight or transparency. This is the current research topic of my team at the Joint Research Centre.

What work are you most proud of (in the field of AI)?

Academically and technically, I am proud of my contributions to music-specific machine learning architectures within the Music Technology Group in Barcelona, ​​which have advanced the state of the art in the field, as is reflected in my citation records. For example, during my thesis, I proposed a data-driven algorithm to extract the tonality of audio signals (e.g. whether a musical piece is in C major or D minor), which became a key reference in the field, and later I co-designed machine learning methods for the automatic description of musical signals in terms of melody (used for example to search for songs by humming), tempo or for modeling emotions in the music. Most of these algorithms are currently integrated into Essentia, an open source library for audio and music analysis, description and synthesis and have been leveraged in many recommender systems.

I am particularly proud of Banda Sonora Vital (LifeSoundTrack), a project awarded the Red Cross Prize for Humanitarian Technologies, in which we developed a personalized music recommendation tool suitable for elderly patients with Alzheimer’s disease . There is also PHENICX, a large project funded by the European Union (EU) that I coordinated on the use of music; and AI to create enriched symphonic musical experiences.

I love the music computing community and was happy to become the first female president of the International Society for Music Information Retrieval, to which I have contributed throughout my career, with a particular interest in music computing. increasing diversity in this field.

Currently, in my role at the Commission, which I joined in 2018 as a Senior Scientist, I provide scientific and technical support to AI policies developed in the EU, including the AI ​​Act. Of this recent work, less visible in terms of publications, I am proud of my humble technical contributions to the AI ​​Act — I say “humble” because as you can imagine, there are a lot of people involved here! As an example, I have contributed to numerous works on harmonization or translation between legal and technical terms (e.g. by proposing definitions based on existing literature) and on the evaluation of practical implementation legal requirements, such as transparency or technical documentation for high demands. risky AI systems, general-purpose AI models, and generative AI.

I am also very proud of my team’s work in support of the EU AI Liability Directive, where we studied, among other things, the particular characteristics that make AI systems inherently risky, such as the lack of causality, opacity, unpredictability or their self- and continual-learning capabilities and assessed the associated difficulties presented when it comes to proving causality.

How can we meet the challenges of a male-dominated technology sector and, by extension, the male-dominated AI sector?

It’s not just about technology: I’m also navigating a male-dominated field of AI research and policy! I don’t have any technique or strategy, because it’s the only environment I know. I don’t know what it would be like to work in a diverse or female-dominated work environment. ” Would not it be nice ? ”, as the Beach Boys song says. I honestly try to avoid frustration and have fun in this difficult scenario, working in a world dominated by very assertive men and enjoying collaborating with excellent women in the field.

What advice would you give to women looking to enter the AI ​​field?

I would tell them two things:

You are indispensable – please enter our field, as there is an urgent need for diversity of visions, approaches and ideas. For example, according to the divinAI project — a project I co-founded on monitoring diversity in the field of AI — only 23% of author names at the International Conference on Machine Learning and 29 % at the Joint International Conference on AI in 2023 were women. , regardless of their gender identity.

You are not alone: ​​there are many women, non-binary colleagues, and male allies in the field, even if we may not be as visible or recognized. Seek them out and get their mentorship and support! In this context, many affinity groups are present in the field of research. For example, when I became president of the International Society for Music Information Retrieval, I was very active in the Women in Music Information Retrieval initiative, pioneering diversity efforts in music computing with a program very successful mentoring.

What are the most pressing issues facing AI as it evolves?

In my opinion, researchers should devote as much effort to the development of AI as to its evaluation, because there is an imbalance today. The research community is so busy advancing the state of the art in terms of AI capabilities and performance and so excited to see its algorithms used in the real world that they forget to to appropriate evaluations, impact analyzes and external audits. The smarter AI systems are, the smarter their evaluations should be. The area of ​​AI evaluation is understudied, leading to many incidents that give AI a bad name, e.g. gender or racial bias present in datasets or algorithms.

What issues should AI users be aware of?

Citizens who use AI-based tools, like chatbots, need to know that AI is not magic. Artificial intelligence is a product of human intelligence. They must learn about the operating principles and limitations of AI algorithms to be able to challenge them and use them responsibly. It is also important that citizens are informed about the quality of AI products, how they are assessed or certified, so they know who they can trust.

What is the best way to develop AI responsibly?

In my opinion, the best way to develop AI products (with good social and environmental impact and in a responsible manner) is to devote the necessary resources to evaluation, social impact assessment and risk mitigation – for example for fundamental rights – before putting an AI system on the market. This benefits businesses and product trust, but also society.

Responsible or trustworthy AI is a way of creating algorithms where aspects such as transparency, fairness, human oversight or social and environmental well-being must be considered from the start of the design process of AI. In this sense, the AI ​​Act not only sets the bar for regulating artificial intelligence worldwide, but it also reflects the European focus on reliability and transparency, enabling innovation while by protecting the rights of citizens. I think this will increase citizens’ trust in the product and technology.