Väitös (tieto- ja viestintätekniikka): MSc Lauri Vasankari

MSc Lauri Vasankari esittää väitöskirjansa ”War Machine Learning - AI in Defence” julkisesti tarkastettavaksi Turun yliopistossa perjantaina 17.4.2026 klo 12.00 (Turun yliopisto, Educarium, EDU 1, Assistentinkatu 5, Turku).

Vastaväittäjänä toimii professori Juha Röning (Oulun yliopisto) ja kustoksena professori Jukka Heikkonen (Turun yliopisto). Tilaisuus on suomenkielinen. Väitöksen alana on tieto- ja viestintätekniikka.

Tiivistelmä väitöstutkimuksesta:

The research looks at the adaptation of Artificial Intelligence (AI) as a technology to modern armed forces and defence through a series of original publications, ranging from high-level systemic perspectives to narrow problem solving with AI algorithms and models.

The key universal findings encountered in the research closely mirror many commercial use cases, albeit with a domain-specific nuances.

Data, not the algorithms or AI models, is the real bottleneck. Across different types of AI, from image recognition to reinforcement learning in decision support, the algorithms fall short due to data scarcity, poor quality, high classification status or siloed efforts.

Hence, the research highlights the need to acknowledge data capability, in other words, the ability to generate and exploit high quality data to actually leverage AI in military use.

The main contribution of the dissertation is the proposed data ecosystem that highlights the importance of operational feedback, data and model governance, evaluation and validation as well as a collaborative framework to bridging silos and stakeholders to solve shared problems with combined efforts.

The research also notes the human role in human-AI teaming, where the premise of AI is not met nor the human operator placed in a respectable position unless the point of view is altered from an overseer to a supporter.

As an impact on the surrounding world, the dissertation calls for smarter, sovereign and focused defence spending on AI, changing the focus of procurement processes to highlight the data readiness that would better enable developing AI capabilities as well as safer human-machine teaming that recognizes human operators as more than just scapegoats for failure.