ARTIFICIAL INTELLIGENCE… ARTIFICIAL LAWYERS? Harry Phillips American Inn of Court January 18, 2018
AI is Everywhere, and its Role is Expanding. Examples Today: Recommendations: Amazon, Netflix, Pandora Spam filters Voice-to-text Facebook tags Nest Thermostats Siri, Alexa, Google Assistant Self-driving cars
How Modern AI Became Possible The dream of intelligent, autonomous machines dates at least to ancient Greece. Three developments have made it possible: More powerful computers Artificial neural networks The use of large data sets
1. More Powerful Computers: Moore’s Law The number of transistors in a dense integrated circuit doubles approximately every two years. This “law” has held for decades:
Moore’s Law: 1969 vs. 2017 Apollo Guidance Computer, 1969 Iphone x CHANGE Processor speed: 0.043 MHz Memory: 64 Kbyte Cost: $3.5 million 2390 MHz 6 cores Up to 256 Gbyte $1,149 336,000 x 4,200,000 x < 1/3,000th https://www.zmescience.com/research/technology/smartphone-power-compared-to-apollo-432/
2. Artificial Neural Networks Traditional, linear programming logic follows pre-specified, human instructions. Artificial neural networks, in contrast, employ countless individually-operative neural switches that are arranged and conditioned to react to inputs en masse.
Traditional programming logic is linear.
Neural Networks use uncountable pathways that are shaped over time in response to data inputs.
3. The use of large data sets: Big Data Modern AI systems are “trained” by running huge amounts of data through a neural network. As the system “learns,” it identifies patterns in the data – patterns not specified by human programmers. This pattern recognition is central to the system’s function: Who likes what movies What is spam What word corresponds to a given sound Whose face is in an image How SCOTUS will rule . . .
Use of AI in Law Today Technology-Assisted Document Review Westlaw Plain Language Searches Contract Review Behavior Prediction (Courts, Jurors . . . )