Peer to Peer Magazine

Fall 2017

The quarterly publication of the International Legal Technology Association

Issue link:

Contents of this Issue


Page 40 of 59

42 PEER TO PEER: THE QUARTERLY MAGAZINE OF ILTA | FALL 2017 Given that AI itself is hard to define, it's also naturally difficult to categorize which technologies constitute AI and which don't. The fact that there are multiple types of technologies that make up the world of AI also adds to the difficulty. "Machines that can read" might sound cuing- edge, but most people deposit checks into an ATM that employs optical character recognition to read the amount on your checks several times a week and don't think twice about it. Same for facial recognition: we take for granted that social media websites can identify our friends in photos. Voice assistants like Siri, Alexa and Cortana probably hit closer to the mark, but the voice prompts you're expected to provide (to be sent off to a server farm for natural language processing that's too slow to run on your home computer) are shorter than a typical conversation starter, and rarely seem very human ("Alexa, open Domino's and ask for my Easy Order"—a real suggestion from the Amazon Echo website). These under-the-hood forms of AI—OCR, facial recognition and natural-language processing done at a server farm somewhere—are probably how AI products will continue to impact daily life. These products suggest a loose working definition of AI: any soware that produces humanlike output (no maer how close to the ball you move the goalposts, and yes, I am looking at you, Alexa). Under this loose definition, AI is already present in products we encounter every day. This is true of technology that's become an integral part of day-to- day legal practice—so much so that any lawyer wouldn't be able to imagine functioning without it. (Whether or not lawyers realize they're using it is a different question.) by Benjamin Whetsell Defining artificial intelligence is hard. One of the most common definitions of AI is the replication of human thought processes—things like learning, decision-making and language—by machines. While this definition isn't wrong, it's circular, and it paints a poor picture of how AI functions on a day-to-day basis in the real world.

Articles in this issue

Archives of this issue

view archives of Peer to Peer Magazine - Fall 2017