Text: Ville Heirola
Iran-born and Turku-highly-educated, Dr. Sepinoud Azimi is a professional in data science and artificial intelligence. Dr. Azimi is currently working as a Senior Lecturer at Åbo Akademi, along with a host of “extra-curricular” activities, such as being an ambassador of the Women in Data Science organisation. With the aim of building more inclusive, fair, and transparent AI applications, her work focuses on exposing the biases of data and AI. Moreover, she is also an international talent in Turku with personal experience of how and what is it like to find oneself in completely new surroundings.
We talked to Dr. Azimi about her experiences, what Turku is like now for international talents, and why her research is crucial for us as AI becomes increasingly ubiquitous. Here’s what we learned.
A Sense of Being Welcome
Like many a good story, Dr. Azimi’s begins with the pursuit of happiness. Leaving Iran was a tough decision, but one that was facilitated by the accessibility and warmth Turku represented.
– I came to Turku in 2009. The reason why I wanted to leave is similar to a lot of people’s: looking for a better life. While many universities accepted me, I chose Turku because the wording in their letter of approval made me feel like they really wanted to have me. That initial sense of welcoming and friendliness was not the case in every country.
Since then, Dr. Azimi has gone on to finish her PhD and landed the position of Senior Lecturer. Yet, it hasn’t always been easy. Fixed-term contracts and employment-based residences can combine to create unneeded stress. The lack of permanency was always at the back of her mind.
– The stress was always there. Building networks and emotional connections is affected by not knowing whether you’re able to stay in the country. The result is an emotional rollercoaster of successes followed by fears. Of course now I am ecstatic to stay here.
Explaining Explainable AI
That rollercoaster of emotions and career twists and turns has brought Dr. Azimi to a position where she does pioneering work in a host of fields that seem, to the untrained eye, wildly different from each other.
– My work has three main tracks: one is health-related, working with medical AI; one with maritime and smart vessels; and one with gender equality. While these may seem different, what is under the hood is the same: applications of AI and data science and finding ways to improve communication.
In fact, much of Dr. Azimi’s work falls under the umbrella of so-called “explainable AI”. Think of it as a kind of meta-AI, which explicates why programmes and algorithms yield the results they do. By examining the programme’s process, explainable AI can build a step-by-step map of the AI’s “thinking”. The kicker is this: explainable AI allows the user to identify any potential bias in the target programme. In an era that is just beginning to lift the veil of objectivity in data and talk about bias in artificial intelligence, Dr. Azimi’s work points the way to a more transparent view of what goes on in the algorithms that decide what we all see, do and respond to.
Unlocking Potential Through Communication
Dr. Azimi’s call for better communication and increased transparency is not limited to algorithms and data sets. It comes down to the people doing the programming. That’s where she would like to see more inclusivity and community instead of isolation and obfuscation.
– People and groups working in AI have a tendency to isolate themselves. I might not know what the person next door is working on, and they might have the perfect solution to a problem I have. Things like the AI Business Academy are great opportunities to alleviate this lack of communication. The terminology and expectations, for example, between healthtech and robotics are different.
Dr. Azimi’s work with explainable AI has as its goal to increase trust and transparency in what can be an esoteric and highly specialised field. In addition to bias and opacity, hype and inflated expectations about AI are another constant nuisance.
– Another point is having too ambitious or too noble ideas about what AI can do. Occasionally, the solution to a company’s problem might not be AI at all, but a bit of statistical analysis.
According to her, Turku has all the knowledge and resources it needs, but more communication between everyone working in AI would not hurt. With increased communication and linkage between different projects, the Turku AI scene can make reality rival the hype.
Opportunities for International Students and Graduates
Through teaching and supervising students at Åbo Akademi, Dr. Azimi is familiar with bias not just in data, but in the real world. The phenomenon of international students leaving Finland due to not finding suitable employment after graduation is a hot topic. With slightly worse luck, it may have happened to her.
– There is a huge amount of potential for students and graduates wanting to build their careers here. However, international talents do face hurdles that others don’t. They need to be seen not as a risk, but as an investment. Of course, I understand that employing someone for a couple of months is not free for the employer either if it doesn’t work out. Every investment has its risks.
In recent years, Dr. Azimi has seen her students’ interests align with the general trends in AI and data science.
– The students we have are fantastic. Machine learning is all the rage right now, and they are gravitating towards that. The possibility of getting hired and having a better salary is high in that segment.
High salaries are one thing. Yet, Dr. Azimi emphasises the overall view of Turku and Finland as more meritocratic and balanced places to build one’s career than many other regions.
– This is the good thing about Finland: If you want to advance in your field, you have a good chance to do that. You might not get paid as much as you would elsewhere, but you don’t spend as much either. Income inequality and unmerited wealth are not as big here. You get a sense of fairness in Finland. You end up feeling like you are where you should be.
