Yara Kyrychenko's PhD explores how social media and social technology can be used to promote pro-social behaviour rather than to stoke divisions.
What I am interested in now is how we can align social technologies like social media and language models to serve human goals and uphold our values instead of perpetuating undesirable social biases.
Yara Kyrychenko
Yaroslava [Yara] Kyrychenko began her PhD in Psychology in 2022, just months after Russia invaded her country Ukraine. It aims to understand how social media, and social technology more generally, can be used to promote more positive societal outcomes rather than to increase division.
A large part of Yara’s work has focused on understanding how people’s social identities and the larger social context interact with digital technologies to produce complex societal outcomes. One example is studying when social media promote solidarity with one’s own group or hostility towards outsiders.
“In general, in political contexts, hostility is more likely to go viral on social media than solidarity, but in certain situations our research shows that solidarity can drive more engagement,” she says. “The algorithms are trying to optimise for people’s attention and you are more likely to pay attention to things that are important to your group in the current social context. What I am interested in now is how we can align social technologies like social media and language models to serve human goals and uphold our values instead of perpetuating undesirable social biases.”
Yara’s PhD centres on social identity biases and how we can measure and mitigate them by drawing on psychological theories and computer science methods. The aim is to promote more nuanced perspectives, to bridge divides without stifling productive conflict and to create scalable pro-social AI. “I hope to make people more aware of the impact they have on AI systems, as research shows that it is often biased human data that leads to biased AI,” she adds.
Early years and undergraduate studies
Yara was born in Kyiv. Her father has just started his PhD in Philosophy and her mother studies Psychology and wants to be a therapist to those fighting on the frontline and those supporting Ukraine beyond the battlefield. Both love learning and conveyed that desire to learn to Yara and her younger brother. Yara’s brother is studying Computer Science and is interested in developing games that move beyond borders to empower people as playing online games with his friends who were geographically scattered after the invasion helped them stay connected. She says she is very proud of her family for their resilience in the last turbulent years since the Russian invasion and for finding something they feel passionately about in the face of adversity.
As a child, Yara was passionate about science and took Maths, Biology, Physics and Chemistry at high school, with Maths being her main focus. In her last year she became a research assistant at the Institute of Physiology of Ukraine and learnt about the possibility of working in research.
Yara started her undergraduate degree at New York University in 2018. She was originally planning to pursue a research career in Computational Biology, but she quickly decided to focus on Maths alone. When Covid hit she wanted to do something to help. “At the time I was really into pure mathematics and thought it was the most beautiful thing, but I also wanted to have a more immediate positive impact,” she says.
By that point she had gathered enough credits and had completed her honours thesis in Maths. She decided to do a second major in Psychology. She was particularly interested in reinforcement learning – a subset of machine learning that allows an AI-driven system to learn through trial and error using feedback from its actions – because of how it links human and machine learning. Yara says she was drawn to Psychology because so many of the world’s pressing problems seemed to have their roots in communication and a failure to resolve differences.
As a young teenager she had witnessed the Maidan revolution of 2014 where people joined together to demand that the Ukrainian president at the time reverse his decision to abandon a political association agreement with the European Union in favour of closer ties to Russia. She saw how peaceful protesters were met with violence yet persevered before the former president eventually fled to Russia, resulting in regime change and a move to greater transparency and freedom of ideas.
In 2021 while completing her joint degree in Maths and Psychology, Yara spent some months abroad at NYU London. It was then that she met her current supervisors. She returned to the US to finish her degree and was in New York when Russia invaded Ukraine. She was deeply moved by how NYU’s Russian society helped organise support for Ukraine and Ukrainians living abroad. Yara stayed in the US for the summer, working as a research assistant at MIT and Cornell before arriving in the UK to start her PhD in Psychology.
Cambridge
Originally, Yara wanted to study the role of social media in promoting collective action. She was fascinated by how social media had seemingly changed from the early 2010s from being a force for uniting people against corrupt or tyrannical leaders to spreading polarisation and misinformation. She was interested in combining her mathematics background with psychological theory to analyse large-scale social media datasets in order to understand how social media could be made more positive and productive.
She has focused her research attention on how conflicts change engagement patterns on social media. Her research has, however, been affected by changes in how Big Tech companies share data so she is looking at alternative ways to explore these questions. She has had to plan around that, for instance, through creating simulated social media feeds to experimentally examine the response to posts that promote hostility and those that promote solidarity in different social contexts. “It is possible that people tend to bond together when under immediate threat or to focus on divisions when things are otherwise going well, and that algorithms pick up on these patterns,” says Yara.
Yara has been working with Professor Sander van der Linden and Jon Roozenbeek at the Social Decision-Making Lab at the University of Cambridge’s Department of Psychology. The lab has been exploring methods for ‘inoculation’ against misinformation, teaching people about the techniques that are used to manipulate them so that they can better discern between what is likely to be true and what is false.
Thanks to the interdisciplinary nature of the lab and the PhD programme at Cambridge, Yara has been able to collaborate with fellow students on a range of issues, from building AI that promotes effective climate action with Miriam Remshard to investigating biases in large language models with Tiancheng Hu, another 2022 Gates Cambridge scholar.
Publications
While she has been at Cambridge Yara has published multiple articles that have received widespread coverage in journals like Nature Communications, Nature Computational Science and PNAS. One study in Nature Communications was described as one of the first major studies of social media behaviour during wartime. It draws on Yara’s research on Ukraine and found that posts celebrating national and cultural unity in a country under attack receive significantly more online engagement than derogatory posts about the aggressors. Yara, who co-organises the Cambridge Social Psychology Seminar Series, is also keen to investigate whether, three years post invasion, feelings of solidarity have ebbed as the war has ground people down, as well as the causal effects of social media on polarisation and pro-group behaviours.
Yara has also recently co-written an article on ‘AI slop’, cheap, low-quality AI-generated clickbait content which often spreads misinformation. The article charts how some social media users are choosing to opt out as a result, potentially further fracturing the public sphere and exacerbating polarisation.
In addition to Yara’s research on social media, she works on the broader problems of Responsible AI. Yara has this year been an Enrichment Student at The Alan Turing Institute of data science and AI, where she has focused on a social evaluation of language models. She has also completed a research internship with Nokia Bell Labs’ Responsible AI team, working on Constitutional AI – a method for training AI systems to be helpful, harmless and honest by following a set of principles, or a “constitution”, instead of relying solely on human feedback which can propagate implicit human biases. For her research and leadership potential, Yara has recently been awarded the 2025 Women of the Future Award in AI.
After she finishes her PhD Yara hopes to work on ensuring advanced AI is safe and beneficial. She plans to investigate the benefits and risks of large-scale AI adoption, Artificial General Intelligence and multi-agent systems, which could have unexpected emergent properties as multiple AI agents interact within a shared environment.
