According to the World Economic Forum’s Global Gender Gap Report 2020, women are particularly under-represented in most of the micro-clusters with the highest employment growth rate that make up the jobs of the future, Data and AI being one of them. Data provided by LinkedIn show that women make up only 26% of positions in the sector, even if this is an improvement over the lower representation in the more established technology professions of Engineering and Cloud Computing.
Per the 2021 Report, the pandemic and world-wide school and childcare establishment closures seem to have affected women’s prospects inversely, by disproportionately increasing their childcare responsibilities and stress levels as compared to men. Flexible working arrangements and remote childcare provisions failed largely at providing a real solution to the issue and made a strong case for the importance of care provisioning for women’s fair labor force participation.
On a different note, enterprises are beginning to understand that they are better served by building diverse teams in terms of gender, race, socioeconomic backgrounds, experience, etc., as this allows them to think differently, innovate and better serve a more diverse client base. The most obvious and easy way to add diversity in the workplace is by ensuring both sexes are equally represented at all levels and there are efforts underway to this end. Enabling the representation of women in AI roles involves more than just focusing on relevant recruitment practices. It has to do with creating an inclusive and flexible work culture that allows women to juggle the multiple roles they often have to fulfill in their lives, especially where children are concerned.
Another way to support women in non-traditional employment roles is by celebrating the ones already there, learning from their experience and listening to their advice. As I have been working with language technologies for several years, on the implementation side of things, I was eager to talk to some women scientists – as opposed to men that I mostly get to work with – and hear their story. I was curious to better understand how and why they chose to go into STEM (Science, Technology, Engineering and Mathematics), what excites them in their work and their thoughts on the workplace gender gap. I have often wondered what my own life would have been like had I chosen STEM as opposed to humanities that I conservatively opted for when, as a teenager myself, I was asked to make a choice of studies. This resulted in a brief chat with three women scientists at AppTek, who work with automatic speech recognition (ASR) and machine translation (MT) technologies and come from diverse backgrounds.
Uma Moothiringote, speech recognition scientist, believes the reason behind the lack of women in AI has a lot to do with a popular misconception that women fall behind in STEM, which makes women be less confident in their abilities. Another popular belief – and a valid one for that matter – is that work-life balance is especially challenging in IT, which may force many women to refrain from entering the industry. Uma was fortunate not to have such an experience: her first job out of university has been to work at AppTek’s Aachen office in Germany, where she studied and her husband also works, and they happily and smoothly transitioned to a family of three recently. So work-life balance doesn’t always need to be bad.
Uma grew up in India and completed her BSc degree in Kerala, and then continued her postgraduate studies in Aachen. She speaks Malayalam, English and Hindi fluently and is currently learning German and Tamil. As a child she loved solving problems and was encouraged to go into science by her close family and teachers alike. She was strongly influenced by her grandfather who was a scientific writer for children and also a translator from English to Malayalam. As a result, Uma is also interested in linguistics, so working on speech recognition seemed like the logical choice to her, as she was able to combine her passion for computers to the love for language her grandfather instilled in her, she says. “It was exciting for me to think that I would be able to do something similar albeit with computers.”
A proponent of social and gender diversity, Uma believes the right mix of people brings about greater innovation and ideas. She advises women that want a career in STEM to work on projects so as to acquire hands-on knowledge and experience, and new skills that will make them stand out in a male-dominated society.
Sarah Beranek, a leading speech recognition scientist at AppTek, thinks increasing the number of women in STEM and their visibility is extremely important if we want to inspire young girls to pursue a career in the field. She encourages women to follow their passion and not to allow the biases society imposes on us from our upbringing affect them. She strongly advises them to talk to other women and share their experience.
A native German who loves traveling, Sarah was influenced in her choice of studies by her physics teacher, who she still remembers fondly. She ended up studying physics in Aachen and Helsinki and then participated in the CMS experiment at the Large Hadron Collider at CERN, in Geneva, during her PhD and in yet another experiment during her postdoc. It was at CERN where she experienced one of the greatest scientific moments in her life in July 2012: The discovery of the Higgs-Boson was announced by the CMS and ATLAS collaboration.
“Being part of the discovery of the Higgs-Boson at CERN and realizing the impact of this discovery for the whole physics community after a 50 year hunt as well as humankind on its path to unraveling the mysteries of the universe was exciting and exhilarating for all of us. ”
There are certainly many different paths to get into the Human Language Technology (HLT) field and Sarah is proof of this. “A strong mathematical background is a must,” she says. She has been working with ASR for the past three years and, instead of reconstructing elementary particles, she now reconstructs the human voice into text, which she says is not that different from a machine learning point of view. She finds ASR a challenging and complex problem with a lot of components. “It is such an active and thriving field - I am constantly learning something new. This I also like the most about my work: it always involves learning and the problems are diverse and challenging.”
Parnia Bahar, a machine learning scientist who combines automatic speech recognition and machine translation expertise, also believes that diversity is needed not only in gender but also in culture, as people with varied backgrounds can offer new ideas and perspectives.
She studied electronics in Stuttgart and then completed a PhD at the human language technology group at RWTH Aachen University, working on alternative models for automatic text-to-text and speech-to-text translation. She gets her passion for computing from her uncle, who was a computer engineer; she learned so much from him growing up that she assembled her first computer at the age of twelve.
Parnia never planned to be a scientist, though she always enjoyed learning new things. She speaks three languages (Persian, English and German) and is excited by all aspects of human language technology, especially its integration with artificial intelligence. “AI seems to bridge the gap between technology and the humanities. As an analytic-type person, I am eager to see how computers can potentially help humans,” she explains. Being on the cutting-edge of developments is of course exhilarating in itself. “The technology is escalating so fast and the industry is very competitive. It needs people with advanced scientific background, comprehensive hands-on-experience, and creativity.”
Parnia currently works on both ASR and MT in an exciting research direction that involves both technologies and is known as speech-to-text translation, i.e. automated translation, straight from audio, that does away with the transcription and text-to-text translation step. She focuses on developing powerful – and also fast – models that can take a speech signal from a source language and translate it into syntactically correct and semantically adequate text in the target language. In doing so she needs to address the challenging aspects of both ASR and MT, which makes the task more complicated than the sum of its parts.
She reiterates the views shared by her colleagues. Passion is important to help one overcome the challenges that the STEM fields are fraught with, irrespective of whether you are a man or a woman, while generous, family-oriented benefits will make a career in AI more viable for women. Gender differences of course do exist, but it is by better understanding our nature and focusing on what each of us has to offer that we can progress successfully as a society. To change things we need more women’s voices to shed light on the opportunities and advantages of the industry.
There is no doubt that STEM fields usually have overwhelming challenges and one needs to be passionate. Never lose sight of why you chose the field in the first place when facing a big problem in your career path. Keep in mind that big challenges occur regardless of gender. If you desire to stay on the edge of novel technologies, think positive, learn new skills, keep motivated without listening to judgments, and continue your absolutely amazing work, which is an inspiration for the next generation of female professionals.
AppTek provides an artificial intelligence and machine learning-based automatic speech recognition, machine translation and natural language understanding platform for organizations in a variety of markets, such as media and entertainment, call centers, government, enterprise business and others across the globe. Available via the cloud or on-premise, AppTek delivers the highest quality real-time streaming and batch speech technology solutions in the industry. Featuring scientists and research engineers who are recognized amongst the best and most experienced in the world, the company’s solutions cover a wide array of languages, dialects, and channels.