Artificial intelligence continues to grow rapidly. Tom Merritt breaks down the five things you need to know about AI, according to a report from Stanford University.
Every year the Human-Centered Artificial Institute at Stanford puts together the Artificial Intelligence Index Report, relying on experts from around the discipline, including folks at Harvard, Google Open AI, and more, to try to pin down where we are with artificial intelligence (AI). You should definitely read all 290 pages, but for now here are five things to know about the state of AI.
SEE: Artificial intelligence ethics policy (TechRepublic Premium)
- People want to learn about AI. AI is now the most popular specialization among computer science graduates in North America at more than 21% of CS PhDs. That’s twice as popular as number two–security.
- Publish, publish, publish. From 1998 to 2018, there was a 300% rise in peer-reviewed papers on AI. China published more AI papers than any other nation. But, you can probably use ML to find the papers you need to cite.
- The US cites, spends, and patents the most. In AI, US authors are cited 40% more than the global average, and we cash in on it. The US leads the world with $12 billion in private AI investment, almost double number two China’s $6.8 billion. And the US files three times as many patents as number two Japan.
- The investment is spread around. Autonomous cars lead the way with just less than 10% of global private investment at $7.7 billion. Medical research and facial recognition are at about $4.7 billion each. But the fastest growing field by dollars is robot process automation (RPA) at $1 billion, followed by supply chain management.
- It’s getting faster (and cheaper) to train algorithms. The time needed to train a machine vision algorithm on the dataset ImageNet took three hours in October 2017; in July 2019, it took just 88 seconds. And it got cheaper too, falling from a few thousand dollars to less than $100.
That’s just where the work is getting done and where the money flows. As far as results, AI seems to be helping make software work a little better. But, most of your human skills are just getting help from the competition, not being replaced… for now.