Stanford AI Index Reveals Exploding Investment and Rapid Training Speeds
The 2019 AI Index report highlights massive global private investment exceeding $70 billion and a dramatic drop in AI training times. The study also shows China matching Europe in AI research publications while AI dominates computer science PhD specializations.
Leaders in the AI community release the 2019 AI Index report, an annual examination of the biggest trends shaping the AI industry, breakthrough research, and AI's impact on society. Compiled by the Stanford Human-Centered AI Institute in collaboration with OpenAI, the report focuses on delivering objective, high-quality data about global AI developments. It covers crucial areas like AI hiring practices, private investment, research contributions by nation, and the migration of researchers from academia to industry.
The report highlights a dramatic reduction in the time and cost required to train AI systems, which historically hinders AI adoption rates. Training a large image classification system on cloud infrastructure now takes about 88 seconds, down from roughly three hours in late 2017. Global private AI investment exceeds $70 billion in 2019, with autonomous vehicles leading the way at $7 billion, followed by drug and cancer research, facial recognition, and fraud detection.
Academic and research trends show explosive growth, as peer-reviewed AI research grows by 300% from 1998 to 2018 and the number of AI papers on arXiv increases 20 times over the past decade. AI is currently the most popular area for computer science PhD specialization, with 21% of 2018 graduates focusing on machine learning or AI. Geographically, China now publishes as many AI papers as Europe, North America claims over 40% of AI conference citations, and nations like Singapore, Brazil, and Canada experience the fastest growth in AI hiring.