Data is Center Stage at AI World 2018

Neurala’s team took on the crowds at AI World 2018 last week, catching up with press, customers and industry peers over the three day conference and expo. With more than 2,200 industry leaders in attendance, it was great to catch up and share what’s new and what’s next in the AI world. From inspection use cases to data annotation and tagging strategies to taking on bias, the sessions were full of exciting insights.

FullSizeRender (4)

Neurala Co-Founder and COO Heather Ames took the podium at two sessions, sharing both her unique perspective on addressing bias and explainability in AI and why edge AI is the future of robotics. A few key takeaways:

  1. Bias continues to be an essential problem in technology communities, AI especially. We almost uniformly see bias as a negative thing, and while it can be in many instances, I’d argue that bias can be positive – for example, expertise bias. We need to understand bias and ensure that if it is present, it adds valuable perspective -- not a slant that favors already privileged groups.
  2. Instead of trying to understand each individual layer of a deep neural network, we should test AI like we interview humans for a job. You wouldn’t give a job candidate an MRI. Based on the data inputs - or experience and education, in the case of the job applicant - consider what outcomes it can produce and why.
  3. The key objective for the robotics community: Capture the human ability to learn day after day. At the cornerstone of this goal is lifelong AI technologies, like Lifelong-DNN™. This approach requires 1000 times less data to train, eliminates catastrophic forgetting, a common pitfall of AI development, and learns instantly without internet connection, keeping data local and private.
  4. Data is the foundation of AI. With data on the forefront of every conversation we had at the conference, it’s clear that organizations that ignore their data strategy are going to face continuous problems deploying AI, both in the near and long term.

Ready to build and tag your data library? Sign up for Brain Builder today.