Editor’s Note:The author of this post is a data scientist at Paul Hastings, and this is the first article in a regular column that will be called “Matters of Opinion.”
There was a time when humanity faced the universe alone and without a friend. Now he has creatures to help him, stronger creatures than himself, more faithful, more useful, and absolutely devoted to him. Mankind is no longer alone.
— Isaac Asimov,I, Robot
By Thomas Barnett, Special Counsel, eDiscovery and Data Scientist at Paul Hastings
In the classic 1950 science fiction bookI, Robotby Isaac Asimov and the 2004 movie named after it, increasingly sophisticated robots eventually clash with their human creators. But in the book, the first battleground is the argument among people about whether using robots to perform humans’ work is even a good idea.
Everyday there are stories about computers orrobots taking on more and more sophisticated tasks, from assisting with financial decisions for global banks, to testing new drugs, to diagnosing illness and even performing medical procedures. Self-driving cars and aircraft are no longer science fiction. IBM’s Watson is marketed as a new generation ofcognitivecomputers that canoutthink our biggest challenges.Are you nervous yet?
[caption id="attachment_5983" align="alignnone” width="640"][Image “Photo by C_osett (Flickr/Public Domain)" (src=https://bol.bna.com/wp-content/uploads/2015/12/17332931280_c9513959f5_z.jpg)]Photo by C_osett (Flickr/Public Domain)[/caption]
In the legal profession discovery documents are classified and contract clauses are analyzed using so-called machine learning, artificial intelligence or predictive coding. But this is just the start. Will advances in computer science lead to technology that can actually do what we think we do better and more efficiently? Will smart robot attorneys be able to assess all the variables in a case and arrive at a precise exposure analysis and settlement value in seconds? Will they be able to take in all of the information in an M&A transaction, conduct due diligence, assess value, risk, and market implications faster than it takes a person to draft an engagement letter?
But what about the basic question, do computers actuallythinkin the way that we do? Or is it something else? Maybe a better, but scarier, question is, how much of what we do as lawyers actually requiresthinkingand how much can be relegated to robots and automation?
Turing is often cited as a proponent of the idea that computers are capable of reasoning and intelligence, but his answer to his own query tells a different story.
[caption id="attachment_5984" align="alignleft” width="168"][Image “Photo of painting of Asimov by thierry ehrmann (Flickr/Creative Commons)" (src=https://bol.bna.com/wp-content/uploads/2015/12/4403111223_5aac04c5af_z.jpg)]Photo of painting of Asimov by thierry ehrmann (Flickr/Creative Commons)[/caption]
As the quote above from the book suggests, this is not a new question. Alan Turing, considered the father of the modern computer, and the subject of a recent movie,The Imitation Game, wrote a paper, interestingly, also published in 1950, the same year asI, Robot, calledComputing Machinery and Intelligence.
The movie is named after the first section of Turing’s paper. He asks the almost unbelievably prescient question, given the state of technology at the time,can machines think? But even though Turing is often cited as a proponent of the idea that computers are capable of reasoning and intelligence, his answer to his own query tells a different story. He suggests that if the question is based on the commonly understood meanings of the wordsmachineandthink, the question isabsurdandtoo meaningless to deserve discussion.
There are a number of scholars who agree. Linguistics luminary Noam Chomsky calls the questionmeaninglessin that we would have to completely redefine what we mean bythinkingto make any sense of the question. Chomsky readily acknowledges that machines can be used effectively to model scientific problems, function as research tools, and help us perform tasks that we find dangerous or unpleasant. But when it comes to understanding and replicating the nature of human intelligence, cognition and thought, Chomsky describes this as a “colossal problem way beyond the limits of contemporary science.” Chomsky goes on to state that it is still unknown whether replicating human thought using computers is beyond the limits of human capability — and he cautions that just because some may desire to do so doesn’t guarantee that it will ever be achievable.
But there are scholars and experts who strongly disagree as the name of the branch of computer scienceartificial intelligence(AI) suggests. Professor Chomsky’s fellow MIT emeriti, Marvin Minsky, considered a father of AI, and Ray Kurzweil, Chief Innovation Officer at Google and the author of the popular book on this topic,The Singularity, both believe that there is nothing fundamentally unique about how the human brain operates, how wethink, that can’t be replicated or even improved upon by computers. They claim that processing electrical impulses in the brain through neural synapses is not that far from what goes on in state-of-the-art computer chips.
But getting beyond the rhetorical debate and the intriguing marketing, how much of what we do as lawyers could potentially be replaced by machines?
Consider this high level description of what lawyers do, theoretically at least, in rendering legal guidance: taking in complex sets of facts and circumstances, considering applicable legal rights and obligations and rendering reasoned opinions and guidance on the best course of action based on all of that information. Built into this highly simplified description is the ability to understand the background and context of events, general knowledge of how the world works and knowledge of the law and its application. Put another way, there is a lot of automatic filtering out of irrelevant noise and focusing in on the signal.
At bottom, computers do one thing and one thing only: switch an electrical impulse on or off.
If you accept this description as reasonable, what if any of that could be done by a computer employing machine learning, artificial intelligence, cognitive programming or whatever term you choose? It is hotly disputed whether even the fastest, most advanced computers could be capable of organizing, sorting and most importantly filtering in or out the appropriate data in order to arrive at a decision.
At bottom, computers do one thing and one thing only: switch an electrical impulse on or off. We translate the on and off into a one or a zero and build upon that single function to perform complex calculations. But computers can do this one thing incredibly fast and incredibly accurately. Far beyond the capability of any human being. If you start from there, you may be able to see how you eventually get to mapping the human genome.
But how about writing Beethoven’s 5th Symphony, painting the Mona Lisa or delivering a quick witted comeback? What about cross examining a key witness in a case? Those examples may not appear as such a clear path from processing a lot of 1’s and 0’s.
There is a vast amount of information about the world that we learn and absorb every second from the time we are born to the time when we are called upon to render complex legal advice.
The standard technology used to understand and interpret potential documentary evidence relies primarily on pattern matching of varying levels of complexity, from matching “key words,” which involves simply seeing if the 1’s and 0’s of one word match those of other words, to so-called “predictive coding,” considered an “advanced” technology in the legal world. As it as currently offered, predictive coding matches the text of entire documents (rather than individual words) to other documents. It is still comparing 1’s and 0’s, but doing it in a more advanced way using statistical sampling and modeling. Work is now being done attempting to use computer technology to analyze damage exposure and likely outcomes based on cases with similar issues, considering the judge, the court, the adverse parties and so on. How close any of that gets to an experienced lawyer’s assessment given all the facts and a career’s worth of experience is another question.
How close or far are we from replicating the magical ingredients of creativity, imagination and innovation? With apologies to Dr. Asimov, so far at least, computers do exactly what they are told, based on highly structured sets of instructions called “programs.” When they go off track it is almost invariably because of an error in the logic of the program — meaning an error by the human programmer.
So far, no one has recorded an instance of a stubborn, insubordinate or subversive computer. Nor, you might argue, is there a creative computer in any sense that we understand the word. There is a vast amount of information about the world that we learn and absorb every second from the time we are born to the time when we are called upon to render complex legal advice. For now at least, finding a computer program that can do that sounds like something from a galaxy far far away.