Could Technology Assisted Review be the answer to saving parties time and cost in discovery?
Key Points
- Artificial intelligence in the form of Technology Assisted Review (TAR), and in particular, predictive coding, is revolutionising the traditionally costly and time consuming process of discovery. Recent Australian cases and practice directions now encourage the use of this new e-discovery technology in civil litigation in Australia.
- Using predictive coding in an appropriate case can save up to 50% on cost, and be 50-75% faster, compared to a traditional fully human manual review.
- Page Seager has recently utilised this cutting-edge technology, reducing an initial 2 million documents to 80,000 that were ultimately discovered, saving significant time and cost.
Discovery – the old way
Discovery is, in general terms, the obligation of each party to a legal proceeding to disclose all documents relevant to the issues in dispute to their opponent. It has come a long way since multitudes of junior lawyers traipsed through rooms full of documents and individually assessed each document for its relevance.
As Vickery J noted in the recent Victorian case of McConnell Dowell Constructors (Aust) Pty Ltd v Santam Ltd 1, if the traditional approach were adopted, “a junior solicitor taking one minute to review and catalogue each of the 1,400,000 documents for relevance in an initial review exercise would take over 23,000 hours, equating to over 583 working weeks. This time does not include the time or cost of a senior solicitor to conduct a review of the process to ensure that only relevant documents are discovered.”
Discovery – the new way
The emergence of e-discovery technologies now allows parties to undertake discovery electronically, in a fraction of the time, with associated cost savings. TAR is used to reduce the number of potentially discoverable documents through keyword searching and concept searching, de-duplication (automated removal of duplicates), email threading (grouping email threads), and more recently, predictive coding.
How does it work?
Recent advances in predictive coding software utilise machine learning to revolutionise the review process. One or two lawyers review and code a subset of documents for relevance, then the predictive coding software learns from the lawyers’ decisions, using computer algorithms to identify and code remaining documents as ‘relevant’ or ‘not relevant’. The process is repeated until a sufficient level of accuracy is reached.
What does this mean for lawyers?
Lawyers will still need to be involved throughout the entire discovery process. Their understanding of the issues in dispute and initial involvement in coding the training set of documents is critical to the success of the predictive coding process. Lawyers are also required to oversee the process, resolve issues and perform quality checks, undertake targeted review of smaller subsets of documents (such as potentially privileged or commercially sensitive documents) and check decisions on relevance and privilege.
What do Australian courts say?
Although it has only been used in a limited number of cases in Australia, in the last two years Australian Courts have begun to encourage the use of TAR, including predictive review, as an acceptable way for parties to complete discovery 2, and, as recently as this year the Supreme Court of Victoria and the Federal Court have released practice notes to that effect.
Page Seager’s experience
Page Seager has recently embraced this newly accepted technology, having used the TAR process in a Victorian Supreme Court case which started with almost 2 million documents (based on relevant custodians and date ranges). This dataset was reduced to 230,000 documents via restrictive search terms, de-duplication and email threading through the Relativity e-discovery platform, and then further reduced to 80,000 documents for discovery using predictive coding. Using this process meant that we were able to complete discovery within the short timeframe set by the court, using a small legal team, while providing a more efficient and cost-effective solution to the client.
Is the use of TAR appropriate for all disputes?
The use of TAR is more likely to be appropriate in cases where:
- there are a large number of electronic documents that require review;
- the majority of documents are electronic and text based. Its value may be reduced if the majority of the documents consist of drawings, plans, images or number based spreadsheets; and
- the size, type and value of the claim are such that TAR would be more time and cost effective than traditional review.
Although figures will vary from case to case, it has been estimated that using TAR in an appropriate case can reduce costs by up to 50%, and can be 50-75% faster than the traditional approach to document review. It can also minimise the interference with a client’s business, by reducing the amount of initial searching and review required by a client.
If you’re interested in detailed advice about whether you could utilise this technology in an upcoming discovery, please contact:
Brett Cassidy
Principal
M: 0438 368 053
E: bcassidy@pageseager.com.au
1 [2016] VSC 734
2 See McConnell Dowell v Santam noted above.
Published: 19 July 2017
