The availability of AI jobs, the capabilities of its systems, and the speed of its deployments are all increasing according to Stanford University’s Human-Centred Artificial Intelligence Institute index. This is a yearly data update that gets issued in association with the McKinsey Global Institute, the world’s No. 1 private-sector think tank. It tracks the growth of intelligence demonstrated by machines across a variety of metrics, from employment numbers through to papers published.

Computer Capacity

Moore’s Law has gone into overdrive according to the Index, with substantial progress being seen in terms of the computing capacity required to run AI being ramped up. Before 2012, results for this sector closely tracked Moore’s Law, but since then it’s been doubling every 3.4 months, an incredible net increase of 300 000X!

Conference Attendance

These numbers have risen sharply, with attendance to AI conferences seeing a significant increase.

In 2019, the Neural Information Processing Systems convention saw 14 000 guests streaming into the Vancouver Convention Centre in Canada, a considerable increase over participant numbers in 2018. Seminars like the Association for the Advancement of Artificial Intelligence and Computer Vision and Pattern Recognition are also enjoying high growth in terms of visitors.

In layman’s terms, it could be likened to the explosion of the eSports industry, one of the fastest-growing industries in the world over the last decade. If you’re not part of the scene yet, the time to learn how easy placing your first eSports bet and winning big is, is now!

Jobs

Another crucial metric is just how many AI-related positions of employment are opening up, one more sector undergoing a boom.

Looking at Indeed, the American worldwide search engine related to recruitment, postings, the share of this kind of occupation has increased five-fold. The fraction of total jobs has risen from 0.26% of all positions posted in 2010 to 1.32% in 2019.

As small a fraction of total jobs as this may be, it’s important to bear in mind that this number includes only positions related to technology working directly in AI development. There is likely an increasingly bigger share of jobs being enhanced and reordered AI.

Training Time

The amount of time that it takes to train AI algorithms accelerated dramatically. It can now happen in almost 1/180th of the time it took a mere two years ago to instruct a big image classification system on a cloud infrastructure.

Back then it would take roughly three hours to teach this kind of system, but by July 2019 it was taking a trifling 88 seconds!

Computer Vision

Image recognition is another benchmark. The Index tracked reporting via ImageNet, which is a public dataset of over 14 million images that was created to tackle the scarcity issue of training data in the Computer Vision field. According to the latest reports, the precision of image recognition by systems has reached an incredible 85%. This is a huge leap from the reported 62% accuracy we enjoyed in 2013.