Artificial Intelligence: Deciphering the Hype vs. the Reality
Following last week’s post on trends on the legal technology horizon, I’d like to take a look at the much-hyped and much-lauded world of artificial intelligence (AI).
Long the domain of science fiction, the term AI conjures up images of robots replacing humans and machines controlling the world. But since IBM’s Watson beat two human Jeopardy! champions five years ago, science fiction has become fact, and the media have proclaimed that the AI era is upon us.
At the recent Code Conference, AI was one of the dominant themes, and companies like IBM, Facebook, Google, Ford, Tesla and Amazon indicated their strong commitment to the technology. When heavyweights like that make a commitment – even if much of AI’s applicability is still speculative – we can’t ignore the potential for it to transform every aspect of our lives, from self-driving cars to smartphones that can read our emotions.
The impact of AI in the legal space is also apparent. There are already a number of companies offering AI-based products and services. For example, machine learning technology is behind the contract analysis platform of Kira Systems, which extract and summarizes contract provisions from large repositories and can learn to find clauses in foreign languages. Catalyst Repository Systems uses a machine learning technology they call continuous active learning (CAL), a predictive coding technique for e-discovery that continuously ranks documents for relevancy and gets smarter with each new batch. Baker & Hostetler, a major law firm in the U.S., recently confirmed that they plan to deploy the first artificial intelligence attorney, ROSS, to help 50 of their lawyers in the firm’s bankruptcy division with legal research chores.
When you think about it, given how much data is in play and how much of it needs to be sorted and analyzed, the legal industry is an ideal testing ground for innovation. We have already seen a number of companies using machine-learning technologies to help lawyers quickly identify relevant documents and cull non-responsive documents and, as I mentioned last week, companies such as Microsoft are tapping into this market to improve document search and management across the enterprise.
So should we prepare for robots to replace lawyers tomorrow? Should we believe that companies like Wavelength – which claims it can unlock “corporate memory” via AI and generate legal playbooks capable of being understood by both humans and machines – will become the norm very soon?
Don’t hold your breath. AI is intended to help lawyers do their jobs, but it cannot instantly replace the intellectual capital and experience of lawyers. After all, AI is based on patterned learning – and in the legal context that means it still relies on feedback from human experts as it refines its decision-making about the meaning or relevance of documents to the substance of a legal matter. Also, AI lacks a lawyer’s instinct, experience and ability to interpret the nuanced human reactions that occur during live proceedings in a courtroom.
That is not to say that things won’t change in the future. There are signs that AI can develop intuition, as we recently saw when AlphaGo – an AI system developed by Google subsidiary DeepMind – beat a Go champion in the highly complex board game that originated in China.
However, for the success of every Watson and AlphaGo, there is a Tay, Microsoft’s AI chatbot that was swiftly terminated when it began using abusive language on Twitter – another reminder that while the potential is enormous, AI is still very much in its infancy.
For now, at least, a robot that can assume the full range of knowledge, skills and judgement required of a competent attorney remains the stuff of science fiction.