Hello..., welcome! I'm Aditya Kaushik, currently a PhD candidate at the ILLC, University of Amsterdam in the Netherlands, working on multimodal event understanding—the models, their behaviors, and the underlying mechanisms.

I obtained my Master's degree in Machine Learning and Artificial Intelligence from Aalto University in Finland.

Prior to that, I worked for 3 years at CommerceIQ.AI and Innova Solutions. I received my Bachelor's degree in Computer Science and Engineering from Amrita University in India.

Contact email: akskuchi@gmail.com

News

  1. 10/2026 Wrapped up a 3-month internship project at Google in Bangalore, India. Many crucial learnings and an overall enriching experience!
  2. 08/2026 Fact, not fib! Our MF2 work, now titled “How Well Do Vision-Language Models Understand Movies?”, is accepted to EMNLP 2026!
  3. 01/2026 Pre-trained multimodal models are increasingly being used in many practical applications involving language generation from temporally-ordered events. But do they have the fundamental ability to recognize salient sub-events in such events? We investigate this and discuss the findings in our latest preprint!
  4. 06/2025 What reasoning skills are needed to understand long videos? Can multimodal AI models comprehend movies like humans do? Or are they biased toward one modality? We explore these questions and release a new benchmark: MF2!
  5. 05/2025 Our work “LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Tasks” is accepted to the ACL 2025 main conference (more details)!
  6. 03/2025 Our work discussing the challenges in evaluating visually-grounded stories is accepted to the Text2Story workshop collocated with ECIR 2025!
  7. 01/2025 New preprint discussing the challenges and future directions regarding “Natural Language Generation from Visual Sequences”!
  8. 11/2024 Here’s the poster I used for presenting our work at EMNLP!
  9. 09/2024 Paper accepted to EMNLP 2024 Findings!
  10. 07/2024 Two new preprints!
  11. 06/2024 Presented work at NLG in the Lowlands workshop in Bielefeld, Germany. Discussed several challenges w.r.t grounding natural language in other modalities and received feedback from amazing researchers!
  12. 03/2024 Attended the ELLIS NLP workshop at the Research Institute for Mathematics (MFO) in Oberwolfach, Germany. Discussed the future directions of LLMs and multimodality with amazing researchers!
  13. 12/2023 Here’s the poster I used to present our work at EMNLP!
  14. 10/2023 Paper accepted to EMNLP 2023 main conference!, presented work at ELLIS, AI4media, and AIDA symposium on large language/foundation models!
  15. 07/2023 Attended the REST-CL retreat in Barcelona, Spain with amazing researchers working on NLP, multimodality, and cognitive science!