The tradition of professionalism is absent in the ABA's recommendations to improve the well-being of lawyers.
The French Government has just banned the use of machine learning to analyze the decisions of French judges. This dreadful law illustrates many things—the growing power of machine learning in legal practice, the centrality of transparency to the rule of law, and the faux liberalism of the government of Emmanuel Macron.
The rise of machines is transforming the practice of law. It has already changed legal discovery. Instead of getting high-paid associates in a room to discover the documents relevant to a case, lawyers now regularly agree on a smart computerized search to find these documents. Computers generate contracts and even simple memos about the state of the law. And machines are beginning to use data analytics to predict the likely outcome of cases.
That latter development should not be a surprise. Famously, baseball scouting was revolutionized when data analytics supplemented the intuitions of old time scouts. Moneyball is coming to law, a field where there is lot more at stake than pitching and hitting.
And importantly once machine learning gets its foot in the door it will keep pushing it open, because machine learning is always improving in scope and even in what must be called creativity. In 1997, IBM beat Gary Kasparov, the best player in the world. In 2012, IBM’s Watson beat the best Jeopardy players in the world. This victory was an advance relevant to law, because while chess is a purely formal system, winning at Jeopardy requires the ability to sort through information in natural language.
In 2017, a new chess machine, called AlphaGo, demonstrated a new level of creativity in machine learning. Big Blue had won because humans programmed the basic goals of chess and then the machine used its superior power to calculate far more positions and look much farther ahead than could even the most expert player. But AlphaGo was just given the rules of chess and played thousands of game against itself during a single day. It then proceeded to beat the best brute force machines in the world, displaying a boldness and a creativity that left grandmasters in awe. This innovation suggests that is quite possible that computers will at some point suggest new kinds of legal strategies.
Just last week, the same artificial intelligence company that created AlphaGo announced that its computers beat excellent human players in multiplayer video games. Again, these computers learned the games by playing many games against themselves. This latest milestone is also relevant to law, because multiplayer video games are games of incomplete information. In other words, a player does not know what his teammate will do but will have to make predictions to decide the optimal strategy. Many legal situations resemble multiplayer games in this respect.
The law banning the application of data analytics to French court decisions would make it harder for people to predict the law. An essential task for lawyers is to make such predictions based on their own study of the law. But lawyers, even teams of them, have limited memory and knowledge. Machine learning can survey and categorize vast amounts of data. It can then take that data and summarize the most relevant to particular judges and situations.
Better legal search and prediction made possible by machine learning have substantial public benefits. As computerized search and prediction became more efficient, people could better plan to fit their conduct to law even if law were relatively complex.
Another great advantage of computerized searches is that it makes law less expensive. Much simple legal search is free on the internet. Poor and middle class people have a lot of unsatisfied legal needs. The rise of computers in law will help meet them.
But despite its benefits, more transparent law has powerful opponents. The foremost of these are members of the bar. Computerized search and prediction are competitors to search and prediction done with more input from lawyers’ labor. Thus, many may be concerned about the competition from machines. To be sure, some will welcome data analytics because they can make use of it and still add their own judgement, making better search and predictions overall. By analogy, at least until recently, while chess computers were stronger than even the strongest chess grandmaster, a chess computer and grandmaster were stronger than the best chess computer. But this development will help only selected lawyers. For instance, superstars will be able to extend their reach by making effective use of machines to do basic work, displacing other lawyers. The rise of superstars has marked the trajectory of technological change in many professions. It helps the earnings of superstars while depressing the earnings of journeymen.
Judges are likely to dislike data analytics as well, because it can show up their divergence from the consensus decisions and indeed may in some cases may suggest they are scofflaws. News reports have thus not surprisingly suggested that French judges may have lobbied for the law banning data analytics
The transparency provided by data analytics can have some potential downsides. For instance, it can make it easier to engage in forum shopping—the practice by which lawyers search for a judge likely sympathetic to their cause. But the only way to stop forum stopping is not to suppress the evidence of its benefits, but to make legal reforms that restrict the practice.
This law provides yet more evidence that the Emmanuel Macron’s government only pretends to be a liberal, reformist enterprise. The rule of law is central to any notion of liberalism—even liberalism very broadly defined. And the transparency of rules is central to the rule of law. One cannot follow an unknown law. Yet here transparency is being suppressed. It is true that one might have an ideal that the legal decisions should not depend on which judge makes the decision, but in a free society citizens should have the right to plan their lives based on the reality that judges matter rather more than the ideal suggests. One problem with dirigiste state like France is that it enforces an ideal of centralized power when the reality is much less ideal than the ideal.
The indifference to liberal values is characteristic of the Macron government. It has just blocked a merger between Fiat and Renault to assure government control of its “champion” in the auto industry. Macron is not breaking the mold, but fits into the cast of a long of French leaders who want to try to control society from the center at the expense of economic and personal freedoms. The right to know the exact content of the law as it actually exists rather than just as it appears on the books is particularly important in a society’s like France where so much is regulated and prohibited.
This French law should prompt Americans to praise our First Amendment freedoms. While we have disagreements about the scope of free speech among ourselves, I do not believe that there is a federal judge in the country who would uphold a similar law in America. Because of the text of the First Amendment and the tradition of freedom it has generated, judges would see the law for what it is—an effort to prevent people from revealing the effects of decisions by their rulers, who emphatically include the judges themselves.