AI Datasets the Key to Better Smart Software
June 5, 2018
The potential for bias in artificial intelligence is a topic on practically every tech junkie’s lips these days. As the world depends more and more on machine learning, we are seeing more clearly the limits and flaws this system possesses. Namely, it’s potential for bias. We learned of this and of solutions in a recent Venture Beat story, “Datasheets Could Be The Solution to Biased AI.”
According to the story:
“Given that many machine learning and deep learning model development efforts use public datasets such as ImageNet or COCO — or private datasets produced by others — it’s important to be able to convey the context, biases, and other material aspects of a training dataset to those interested in using it.”
These datasets are proving to be incredibly valuable and an incredible way to remove the bias from AI. For example, datasets have been used to detect abnormalities in CT scans using artificial intelligence. In fact, they detect the problematic areas at incredible speeds. According to the story: “Qure.ai seeks to speed up the process by leveraging artificial intelligence to screen CT scans in under 10 seconds to find any abnormalities.” Wow, that’s an incredible leap. When AI is given an appropriate foundation like that, the potential for success is off the charts.
Patrick Roland, June 5, 2018
Comments
One Response to “AI Datasets the Key to Better Smart Software”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
Not yet, but I