Apple's Kept Promise: Open Source AI Studies

Apple promised to start publishing some of its AI research, and the first study made public has hit scientific journals. Could this be the move Apple needs to secure their spot in the AI market?

A few months ago Apple promised to publish a research paper about artificial intelligence, essentially creating their open source AI, and they have made good on that pledge. Researcher Ashish Shrivastava and his colleagues submitted their paper, “Learning from Simulated and Unsupervised Images through Adversarial Training,” on November 15, 2016, and it was published on December 22nd through the Cornell University Library.

Apple is taking a risk by making their research public, but it may pay off in the battle for the top of the AI marketplace. By allowing their talent to publish their discoveries, new talent may see Apple as a more attractive employer, and that may just give the tech giant the edge it needs.

The Long Awaited Open Source AI Research

The paper mainly focuses on how machines can use unsupervised learning to study simulated images. This ‘Simulated+Unsupervised (S+U)” learning method allows algorithms to improve their ability to discern what they see when presented with a picture.

So far, simulated images are easier for AI to process because they are usually labeled, but this may not translate over to real-world images which lack pre-prepared labels. Apple notes that relying entirely on simulated images may yield bad results. Yet, S+U learning helps to bridge the gap by giving AI something to cross-reference with unlabeled, real-world pictures.

Sounds Interesting, but is it Actually Useful?

While the open source AI study does give some interesting tidbits about helping AI judge images, it has implications of moving beyond static pictures, and into videos, however, it doesn’t mean much to the consumer at this stage.

The topic of the research may not be the most exciting subject, but it represents a larger move by Apple to be more open to the public about its technological breakthroughs. Previously, Apple had a ban on their researchers publishing findings, and the lifting of that ban may be the most useful thing about the new study.

If new researchers flock to Apple because it supports their findings, then the future of AI research is in Apple’s hands. By leveraging their research as a lure for the scientific community, Apple has a firm stance from which to stay in the fight for the commercial AI market.