No matter how much technology accelerates cumbersome processes like document review, the human element must always be hovering in the background. When it comes to e-discovery, there are five essential stages to the process: find, collect, review, redact and produce. The further up the chain, the greater the need for human judgement and sign-off.
This may reassure lawyers feeling threatened by the rapid proliferation of legal technology, but it comes with one obvious drawback. A carefully crafted algorithm can be undone in seconds by that oldest of human failings: error. Ultimately, nothing is foolproof. ‘This isn’t a situation where robots will replace the attorneys,’ says Dechert partner and e-discovery specialist, Ben Barnett. ‘Even with all the advances in machine learning, the algorithms are predicting relevance based on human input (coding). You still need smart people who can distinguish a smoking gun document from one that is merely relevant and current programs can’t do that. The potential pitfalls from over-reliance on machine learning run the gamut from missing key documents to production of privileged documents.’
