As anticipated by experts, 2017 was a remarkable year for Artificial Intelligence as interest by industry and government intensified and funding for research increased. Here’s a recap on where AI went this year:
Driverless Cars. A Rand Corporation study released in November warns that a delay in allowing autonomous vehicles (AVs) on the road may result in thousands of needless deaths. AV advocates argue that as soon as machine drivers perform even marginally better than humans, they should be allowed to replace humans in order to statistically reduce crashes. Heeding this reasoning, Congress unanimously approved legislation to streamline the introduction of AVs onto the nation’s freeways.
Medical Diagnosis. A scientific paper published in November announces that an algorithm called CheXNet ” . . . can detect pneumonia from chest X-rays at a level exceeding practicing radiologists.” The paper reports that average human radiologists were bested by AI in specificity and sensitivity. Meanwhile, AI is helping human diagnosticians keep their edge through crowdsourcing via the smartphone app, Human Dx.
Factories. Though Taiwanese electronics manufacturer Foxconn committed in 2010 to deploy one million robots in its factories, it has only deployed 40,000 so far. To accelerate the transition, Foxconn is teaming with Landing.ai to make better use of advances in artificial intelligence. Machine learning will be implemented for such labor-intensive tasks as visual inspection of printed circuit boards and adjusting the configuration of injection-molding equipment.
Military. Slaughterbots, a short video released by the Future of Life Institute, portrayed swarms of AI-piloted mini-drones dispatched to assassinate specific individuals, targeting their victims by utilizing pattern recognition algorithms to identify their faces. This isn’t far from present-day reality, as already the military is testing autonomous drone swarms released by fighter jets on simulated anti-insurgency missions. Stuart Russell, a computer science professor at Berkeley University, warns that the window is “closing fast” for banning the spread of autonomous killer robots.
Space Exploration. In August, participants at NASA’s Frontier Development Laboratory showcased an algorithm used to visually identify meteor showers in order to locate undiscovered comets; the AI agrees with human observers 90% of the time. This is outdone by a neural net that is examining lunar photos to classify craters; it agrees with human experts 98% of the time.
Clearly there are many developments that arose in 2017, and the buzz for artificial intelligence is sure to only get louder in the new year. Keep an eye on my blog for new updates on the industry and more history on AI tech.