Making AI Accessible

There is an explosion of research in artificial intelligence (AI) with new methods and models emerging all the time. The software tools for creating, running and training models as well as data is often freely available online. Still there is a barrier that hinders many end users of AI to access and use the latest results. Often an extensive knowledge in programming is needed to download and use the online repositories containing the AI models. In many cases, being a good programming doesn’t even help since crucial components such as data or instructions may be missing in the publications or the associated repositories.

This talk is about lowering the thresholds to AI for those of us that like to use it. I will talk about what one could do to make AI research more accessible and reproducible and what the current status is in that regard. Finally, I will talk about a software we have developed as a spin-off from our research in medical image processing with the goal of making advanced image processing more accessible. In the software we want AI models to be only a few clicks away from being integrated in any image processing pipeline. In this way we want to eliminate the barriers keeping people from testing and using the latest AI models.