I recently tweeted a link to the article Will Computer Learning Replace Lawyer Earning? In many ways it was a typical Artificial-Intelligence-and-the-law article, albeit authored by a data scientist rather than a wannabe lawyer-technologist. Though I can now confirm that the author, Thomas Barnett of Paul Hastings, also has a law degree, so maybe I oversold the data-science angle a bit.
Even so, I was surprised that my tweet of this article was liked and re-tweeted much more than usual. Which led me to re-read it to see what was catching people’s attention. Here’s my pick of the most salient/interesting points:
1. How Much Do Lawyers Actually Think? The AI fallacy describes the false assumption that technology in the future will do our jobs only insofar as it is capable of thinking like us. Barnett broaches this issue briefly, but then somewhat surprisingly asks “how much of what we do as lawyers actually requires thinkingand how much can be relegated to robots and automation?”
2. Replicating Human Thought: Alan Turing considered it absurd and meaningless to ask if machines can think using the common meanings of these words. Noam Chomsky similarly questions whether computers can replicate human thought. Yet Marvin Minsky and Ray Kurzweil believe that computers are not that far away from mimicking neural synapses in a substantively meaningful way.
3. Still just Zeros and Ones: Barnett writes: “At bottom, computers do one thing and one thing only: switch an electrical impulse on or off.” It is their speed and accuracy that’s astounding. But they still need to be told what to do, and even predictive coding can’t really capture “the magical ingredients of creativity, imagination and innovation” which depend on a life-time’s worth of experience.
4. Not Yet Truly Human: Computers with human attributes such as stubbornness and insubordinate behaviour are yet to be created. AI that can match the depth of knowledge and insights we accumulate from the day we’re born to the day we die – the kind of knowledge and insights needed to provide complex legal advice for instance – is still something that only exists in science fiction (specifically Star Wars, according to Barnett).
Broken down like this, Barnett’s article arguably doesn’t add that much to the current debate over whether AI will soon be taking over lawyers’ jobs. And I don’t know whether he actually answered the question posed in the title to his article, to which I guess the answer is “No.”**
Still, his reminding us that computers are just machines that process 1’s and 0’s really fast is a useful takeaway. That we still have a distance to go before truly “thinking” computers arrive also cuts through the hype that bubbles up from time to time within the AI and LegalTech communities.
After re-reading Barnett’s article, I’m personally more convinced than ever that focusing on how we “think” is not as useful as focusing on what we “do” – and then asking whether technology could do the same thing perhaps by different means. Which is why I found another article recently published by McKinsey & Co to be a more compelling read than Barnett’s – although I did think his was a nicely grounded approach that highlighted issues too often glossed-over.
So why didn’t my tweet of the McKinsey article titled Four Fundamentals of Workplace Automation become as popular as my tweet of Barnett’s article? Perhaps it was because the latter focused on lawyers (a topic of particular interest to my twitter followers) while the former maintained a broader perspective, albeit with some references to lawyers in the middle. Or maybe it was because Barnett’s article was particularly well-written and was perceived to be a thoughtful piece by a broader range of readers. Or it could simply have been the randomness of tweeting itself.
What do you think?
ABOUT THE AUTHOR
Peter Macmillan (@pjmcls) is a leading authority on how legal experts think. A lawyer and cognitive psychologist, he has pioneered new empirically-based approaches to the analysis and measurement of specialist legal expertise. He is the inventor of the Calepr Rating Scale – a diagnostic tool for identifying a lawyer’s level of expertise – and author of the book Unlocking the Secrets of Legal Genius: Measuring Specialist Legal Expertise Through Think-Aloud Verbal Protocol Analysis.
NOTE: My commentaries focus primarily – and often exclusively – on the cognitive skills of legal specialists working in top-tier law firms (aka BigLaw) and in highly-regarded boutique law practices. My principal area of interest is the execution of complex, non-routine legal work. The above analysis might not translate to other areas or aspects of the legal services industry.
** Betteridge’s Law: “Any headline which ends in a question mark can be answered by the word No.”