Do Lawyers Have Artificial Intelligence?
Artificial Intelligence (AI) is almost here. Lawyers rely on knowledge and intelligence in practicing law. Technology has made lawyering more difficult by increasing the amount of data that represents potential evidence while making it easier to search, retrieve and review such data. AI represents the potential to better organize the ever-increasing amount of data and make "finding" data easier. Implications for Law? Many. Below is just an overview of some.
Most readers of this blog are familiar with the name Richard Susskind. His most recent book “The End of Lawyers
” has been a fixture in the legal transformation debate and has slowly penetrated the everyday lexicon of the legal practice – though there are still many lawyers who have not heard of him. The name Watson – most people will recall as Arthur Conon Doyle’s fictitious sidekick to Sherlock Holmes. There is a new Watson however and like the literary character this one could put Holmes and any of us through our paces in terms of reasoning and having the answers.
Watson is an IBM supercomputer that was built to advance the theory that computers can and will be able to “understand natural human language.” Note: the computer was named after IBM founder Thomas Watson not the literary figure. Its predecessor, Deep Blue, rose to fame by beating the Chess World Champion in 1997. Watson took what Deep Blue was many steps further in that it can listen to/read language as spoken by humans and respond accordingly. To do this requires amazing amounts of processing power and storage. Watson must be able to detect nuance, puns, emotive language, whether a statement is a declarative or a question, and overall – understand what is a person “saying.” Its processing power is an astounding 80 trillion operators a second which allows it to scan its brain (consisting of 200 million pages of content) in 3 seconds. This brain was developed by scanning in everything from encyclopedias and books to move scripts. Watson is not small – it is roughly equivalent to 10 refrigerators. This bring to mind the first computers ever built that filled rooms and had roughly the computing power of a throw-away solar powered calculator that sits on many of our desks today.
Watson is powerful to be sure - so powerful that it can compete against humans on the game show Jeopardy!. And it wins! The show proceeds like any other where there are three human contestants expect in this case the show takes place where Watson is located. The host reads the answers as he would on any other show and the contests are to “buzz in” and respond with the correct question. Watson participants just like any other contestant but for in place of a human body there is the logo for Watson – behind which is Watson itself.
Moore’s Law + Kryder’s Law = AI
Most people are familiar with Moore’s Law which basically states that the number of transistors that can be placed on an integrated circuit doubles every 18-24 months – doubling the power and memory of a microchip. A lesser known but equally critical law (though debated as to being an actual law) is that of Mark Kryder which states that the amount of information that can put placed on a hard drive doubles every 13 months. Later research has found that it is more close to 18 months. Regardless, the trends on both fronts are advancing and so we are getting much faster computers able to store much more information. Watson is the current barometer for this potent combination. While today it sits inside a room and has an enormous foot print – so did the world’s first computers. How long will it take the capabilities of Watson to be delivered in a computer the size of a pocket calculator (or an iPad) – or even just in a more manageable size of a typical rack server?
In Susskind’s 1999 book “The Future of Law
” he briefly touches upon some implications for the law for when artificial intelligence becomes a reality. “The shift from data processing, through information processing on to knowledge processing is all about increasing performance, range, and scope of computers, moving technology from mere number crunching and data storage and retrieval to task which would depend on knowledge and intelligence” Susskind writes. Knowledge and intelligence has been “the line in the sand” separating humans from computers since the dawn of the technical age. If computers are to step over this line, the future of law (and other professions e.g. health-care) will be here. In other words we will then witness artificial intelligence and are on the brink of convergence (or The Singularity). In terms of AI’s impact on the law, Susskind uses the example of computerized judges. Making the distinction between matters that involve moral judgment (something Susskind recognizes may be the next challenge for computing) and those that involve mere factual issues. He states that there is some “promise of systems which will help choose between diverging accounts of facts of cases – by applying probability theory together with the rules of evidence.”
Obviously another potent use of this type of AI would be to make ediscovery as simple as collecting data and then literally asking the computer to find all relevant information. Actually if we had AI, ediscovery would be easily eradicated. The implications on data management for any organization would be that business intelligence, risk management, compliance, knowledge management, information governance, and records management are made to be so easy and intuitive that there would be no need for ediscovery as we know it today. I predict that we would look back on ediscovery and think of it similar to how we view VCRs in today’s iTunes on-demand Netflix smartphone world – clunky and redundant. Nick Brestoff wrote a piece in Legal Technology News earlier this week that looks specifically at AI and ediscovery. Nick draws a similar conclusion as I have - the true power from a legal perspective of AI would lie not with making dispute resolution easier but rather making dispute-avoidance a reality.
If this all seems too distant or too nerdy-techy to resonate with lawyers, I invite them to watch Watson on Jeopardy! on February 14 through 16 via their iPad (or Slingbox). They will be watching "the future" now on a device that would have been as crazy an idea back in the VCRs days as AI may appear to be today. Changes in the legal profession are coming. Law has already become more technological (ediscovery is just one example). Lawyers will continue to further interface with technology – technology that is gaining in knowledge and intelligence.
Most readers of this blog are familiar with the name Richard Susskind. His most recent book “The End of Lawyers
” has been a fixture in the legal transformation debate and has slowly penetrated the everyday lexicon of the legal practice – though there are still many lawyers who have not heard of him. The name Watson – most people will recall as Arthur Conon Doyle’s fictitious sidekick to Sherlock Holmes. There is a new Watson however and like the literary character this one could put Holmes and any of us through our paces in terms of reasoning and having the answers. Watson is powerful to be sure - so powerful that it can compete against humans on the game show Jeopardy!. And it wins! The show proceeds like any other where there are three human contestants expect in this case the show takes place where Watson is located. The host reads the answers as he would on any other show and the contests are to “buzz in” and respond with the correct question. Watson participants just like any other contestant but for in place of a human body there is the logo for Watson – behind which is Watson itself.
Moore’s Law + Kryder’s Law = AI
Most people are familiar with Moore’s Law which basically states that the number of transistors that can be placed on an integrated circuit doubles every 18-24 months – doubling the power and memory of a microchip. A lesser known but equally critical law (though debated as to being an actual law) is that of Mark Kryder which states that the amount of information that can put placed on a hard drive doubles every 13 months. Later research has found that it is more close to 18 months. Regardless, the trends on both fronts are advancing and so we are getting much faster computers able to store much more information. Watson is the current barometer for this potent combination. While today it sits inside a room and has an enormous foot print – so did the world’s first computers. How long will it take the capabilities of Watson to be delivered in a computer the size of a pocket calculator (or an iPad) – or even just in a more manageable size of a typical rack server?
In Susskind’s 1999 book “The Future of Law
” he briefly touches upon some implications for the law for when artificial intelligence becomes a reality. “The shift from data processing, through information processing on to knowledge processing is all about increasing performance, range, and scope of computers, moving technology from mere number crunching and data storage and retrieval to task which would depend on knowledge and intelligence” Susskind writes. Knowledge and intelligence has been “the line in the sand” separating humans from computers since the dawn of the technical age. If computers are to step over this line, the future of law (and other professions e.g. health-care) will be here. In other words we will then witness artificial intelligence and are on the brink of convergence (or The Singularity). In terms of AI’s impact on the law, Susskind uses the example of computerized judges. Making the distinction between matters that involve moral judgment (something Susskind recognizes may be the next challenge for computing) and those that involve mere factual issues. He states that there is some “promise of systems which will help choose between diverging accounts of facts of cases – by applying probability theory together with the rules of evidence.” Obviously another potent use of this type of AI would be to make ediscovery as simple as collecting data and then literally asking the computer to find all relevant information. Actually if we had AI, ediscovery would be easily eradicated. The implications on data management for any organization would be that business intelligence, risk management, compliance, knowledge management, information governance, and records management are made to be so easy and intuitive that there would be no need for ediscovery as we know it today. I predict that we would look back on ediscovery and think of it similar to how we view VCRs in today’s iTunes on-demand Netflix smartphone world – clunky and redundant. Nick Brestoff wrote a piece in Legal Technology News earlier this week that looks specifically at AI and ediscovery. Nick draws a similar conclusion as I have - the true power from a legal perspective of AI would lie not with making dispute resolution easier but rather making dispute-avoidance a reality.
The Future is Now!
If this all seems too distant or too nerdy-techy to resonate with lawyers, I invite them to watch Watson on Jeopardy! on February 14 through 16 via their iPad (or Slingbox). They will be watching "the future" now on a device that would have been as crazy an idea back in the VCRs days as AI may appear to be today. Changes in the legal profession are coming. Law has already become more technological (ediscovery is just one example). Lawyers will continue to further interface with technology – technology that is gaining in knowledge and intelligence.