Iris.ai the Scientist
— your invaluable team member
Iris.ai is training to become a fully fledged researcher, doing literature based discovery. She will make connections between millions of research based hypotheses to build and validate new ones.
Iris.ai – the scientist
Today’s available Iris.ai tools are a small beginning. Ultimately Iris.ai will be an invaluable companion to researchers, with the ability of doing inference and find new solutions from a vast body of scientific knowledge.
- Summarize a paper’s core hypothesis
- Allow exploration of e.g. only methods
- Connect individual components to another research
- Highlight contradictions with other findings
- Suggest new hypotheses for you
- Validate new hypotheses and publish
The tech required to build an AI Scientist
Still almost a decade away, there is a range of research challenges required to overcome in order to build an interdisciplinary AI Researcher.
Pseudo-hypothesis extraction
Build a graph representation of the input document, linking words from each class (problem, solution, evaluation, result) to words from the other classes forming graph path of the form “problem – solution – evaluation – result”.
Truth tree
Unify all individual resources into a graph representation with the true meaning of each document: a conclusion, a hypothesis, and/or a logical statement, which can be represented in a text or mathematical expression, and subsequently validated or refuted.
Build new knowledge
Do inference on the existing knowledge in order to discover knowledge gaps/white space as well as new knowledge to be implemented.