Introducing AerospaceAI’s Research Engine.

Once upon a time Isaac Newton sat under a tree. It would take only a single observation, the apple
falling on his head, to come to a conclusion and make a monumental discovery about the nature of
physics. Today, a research paper on average will hold a thousand observations, or datapoints,
before arriving at a conclusion that may have similar implications, and every year some 5000 papers
will be published in the aerospace industry alone.

It is not hard to argue that in the huge amount of data and discoveries some findings will be
overlooked or even worse, never get made in the first place. Yet the process of comparing all the
outcomes and rate them accordingly is something we can not possibly expect any human to be
capable of doing. However it is something we can expect a machine to do. This is what we set out to
accomplish with AerospaceAI.

Over the years we have worked to create an assistant, albeit a digital one, that helps to define and
refine the researcher’s scope and work. It takes the available research, understands its relevancy
and context, and presents it to the researcher in a concise manner with actionable ideas. For the
researcher, the whole body of knowledge in a respective area becomes a more tangible and
graspable concept.

The AerospaceAI engine has been trained on more than 10,000 papers and a million data points to
be able to understand relevancy, context, and quality. These traits are also its most experimental
ones, as deep learning models are not yet entirely proficient at understanding human language in all
its aspects. It then proceeds to do what machine learning do exceed in; is to dissect, clean up and
recognize patterns in data sets and return relevant findings and predictions.

In a more practical sense this implies that the user can verify his or her research question quickly
while keeping oversight of the available research throughout the process. It is our hope this will free
the mind of the researcher and allow it to do what it does best; to explore and wonder. This will
lead to more diverse research and improve the quality of the subsequent findings. In the long term
it may provide greater leeway in exploring the more challenging problems we face in the industry.

(as published in the Aerospace Europe Bulletin, August 2021)