The Limits of Technology—and Why We Still Need People

There’s no doubt about it: artificial intelligence (AI) has made amazing advances. AI systems had already mastered Jeopardy! and chess, and now another AI has bested the human champion in the complex game of Go. AI isn’t just playing games, either. If you want to drop $16,000 for a holiday gift, there are robots powered by AI that can independently assess, sort, and fold disorganized laundry.

The same progress is occurring in legal technology, where AI applications are making eDiscovery both easier and—unlike home laundry robots—less expensive.

But technology isn’t without its limits. Sometimes “smart” technology is anything but smart! That’s why you still need a team of qualified, experienced people backing your technology.


What Legal Technology Can Do …

AI is continually improving its understanding of language. With natural language processing technology, AI can now complete a host of data-processing tasks. For example, email threading, near-deduplication, and concept clustering all rely on computers’ ability to recognize similar documents and group them together. These approaches substantially reduce the volume of data for downstream eDiscovery review and production, speeding up the process while reducing costs.

Nor does the usefulness of AI stop after data has been processed: one of the earliest and best-known eDiscovery applications of AI is in technology-assisted review (TAR). Modern TAR systems use machine learning to literally “watch and learn” from human review teams. These programs create their own rules for which documents are most likely to be relevant and filter those documents to the top for review, replacing slow and costly linear review methods.

In short, AI—like other forms of automation in legal technology—excels in managing vast quantities of disorganized data. It can sort and categorize information, spot patterns, and learn from experience, all of which makes it look tremendously smart.

Unfortunately, sometimes appearances can be deceiving.


… And What Legal Technology Can’t Do (at Least Not Well)

AI in general can look as if it is understanding language and concepts when, in reality, it isn’t. Just because a computer can recognize words that tend to be found together doesn’t mean it truly grasps what those words mean. As Melanie Mitchell, a computer science professor, recently opined in the New York Times, “Anyone who works with A.I. systems knows that behind the facade of humanlike visual abilities, linguistic fluency and game-playing prowess, these programs do not—in any humanlike way—understand the inputs they process or the outputs they produce.” This has made AI systems vulnerable to hacking, manipulation, and misuse, sometimes in simple ways that a human would immediately detect.

This underscores the basic fact that computers, for all their capabilities, find certain human functions—walking, getting dressed, recognizing the concept of a “cat”—stunningly difficult. We don’t think that means that computers are inept or incapable; rather, we think it highlights the incredible power of the human mind.

Humans, for example, can weigh conflicting values and make ethical decisions that comport with the most important of those values. AI systems are not even aware of values; they’re perfectly happy to cheat if it helps them achieve their stated objectives. In legal technology, that means that AI systems will not recognize the “flexibility and subtlety of legal language [that forms the] prerequisite for a just and accountable social order”—at least not anytime soon.

So does this mean we should stop using or trusting AI applications in eDiscovery? Not at all. It only means that we should remember the importance of the human support team, which fills the gaps left by technology.


People Bridge the Gap

Unless and until we experience a quantum leap in our understanding of human intelligence and our translation of that intelligence into machine applications, we will need lawyers and technologists to bridge the gap. Technology can readily identify patterns; humans need to be standing in the on-deck circle to determine what those patterns mean in this case, for this client, under these facts.

But lawyers should, in turn, recognize the importance of technology and embrace technologists as an integral part of their team. Prompt, cost-effective eDiscovery increasingly relies on advances in legal technology, supported by software engineers, technology vendors, and AI developers.

Are you ready to streamline your eDiscovery through smart technology that’s backed by expert human support? We’ve been leading the field in both eDiscovery technology and outstanding support for years. Start a conversation with us today to learn more.

Reader Interactions

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.