Communications of the ACM
AI Is Learning From Humans. Many Humans.
Artificial intelligence systems must learn from troves of data labelled by humans before they can be used for their intended purposes.
Credit: Rebecca Conway/The New York Times
While artificial intelligence (AI) systems are viewed by most people in the technology field as the future of the industry, tech executives rarely discuss the labor-intensive process that goes into creating these new programs.
AI systems must learn from human workers labeling troves of data before they can effectively be used for their intended purposes.
Tens of thousands of workers spread around the world, many of them independent contractors working from home, annotate data through crowdsourcing services like Amazon Mechanical Turk.
The market for data labeling, which accounts for about 80% of the time spent building AI systems, passed $500 million in 2018, and will reach $1.2 billion by 2023, according to the research firm Cognilytica.
While AI researchers hope they can create system that can learn from smaller amounts of data, human labor will be essential to the industry for the foreseeable future.
From The New York Times
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found