AI innovation long-term solves the DA.

Given the legal focus of this volume, this section focuses on the benefits to the efficiency of legal services and the broadening of access to these services. AI is involved in improving efficiency at every stage of law, including the discovery in litigation, search for relevant cases, generation of transactional documents, and prediction of legal outcomes.22 To summarize: High-level research in computational models of legal reasoning that when combined with the greater capacity to extract information from legal documents will lead to more autonomous forms of legal reasoning. By lowering the cost of finding and litigating about the law, these changes will not only make the law more efficient but also improve access to justice. While the organized bar has various rules that attempt to prevent many of these developments, it will not succeed in preventing these innovations from making legal services more efficient.23

Machine intelligence is most advanced in legal discovery.24 Predictive coding is the practice by which lawyers look at a sample set of documents discoverable in a case. Algorithms are then constructed to predict which documents are relevant. Predictive coding has transformed legal discovery with the law firms setting up e-discovery units within their firms and independent businesses offering innovative services. It has been estimated that this industry is already worth US$10 billion. It improves efficiency by radically reducing the time that lawyers must devote to searching for relevant documents.

Computerized search has been available in rudimentary form since the 1960s.25 But it is improving as computation improves.26 First, search is improving in finding cases by algorithms that are better at finding the relevant cases. Second, firms are using network analysis to assess which cases are the most important. Third, cases may be connected to the legal briefs that cite cases, better assessing the value of citing particular cases based on outcomes. While much of commercial search still depends on keywords, there has been progress in so-called semantic search, where legal documents are tagged to reflect higher-level semantic meaning.

AI will also improve the production of transaction documents.27 It is already well known that firms like Legal Zoom have generated forms for wills and trusts, broadening access to rudimentary legal services.28 But other firms are using computers to produce more complex documents for a wider variety of transactions. One advantage of such production is that the differences between varieties of transactional documents will be able to be traced and their results in litigation assessed.

AI will also improve predictions in the legal field.29 Much of lawyering consists of predicting outcomes, like whether a patent is valid. AI can use big data to help make such predictions, increasing efficiency and reducing likely litigation, because parties will be more likely to settle if they can agree on a predicted outcome. Already, companies are offering predictive services in discrete areas of the law, like patents.30 As AI improves, it should be able to offer these services in more legal areas. Efficient predictions of case outcomes would broaden access to law.

Ultimately, the holy grail of the intersection of AI and law is the creation of machines that can engage in reasoning, not just predict the reasoning of others. Progress is being made here as well. Programs like Watson allow for information extraction from legal texts. This information can then be placed within computational models of legal reasoning to yield legal argument based on facts and circumstances.31