Posted by Carole Marsden, chief commercial officer at Legal Futures Associate poweredbypie
Marsden: Wading in with AI on a large scale will fail
“Aim for simplicity in data science. Real creativity won’t make things more complex. Instead, it will simplify them” – Damian Duffy Mingle
The very best technology allows staff simply to get on with their job. It works away in the background and we don’t even notice it’s there. There are, in fact, probably many instances where everyday processes are taking advantage of artificial intelligence (AI) and we don’t even realise.
In the near future, we anticipate machine learning will start to be used in the conveyancing process too. It will be used to check documents and recognise images with the aim of speeding up the overall workflow.
In fact, AI machine learning can be trained to establish patterns in data better than a human can, which can be used to drive interesting insights and make better strategic decisions.
However, just as large-scale, end-to-end IT implementations designed to solve total workflow processes often result in an unsatisfactory outcome, wading in with AI on a large scale will also fail.
The best way forward is instead to tackle and safely approach individual elements of the conveyancing process, gradually over time. In this way, staff will learn to trust technology and be able to unlock the advantages of machine learning.
We currently see two immediate areas that are of most relevance to solicitors: cognitive services, which is where the computer learns to recognise images; and AI decision-making, where the technology can make intelligent decisions based on data.
One application of cognitive services is file plan validation. Around 10% of addresses provided by solicitors for property searches are not standard. This may be because the property is a new build or a parcel of land.
My personal favourite was described by one law firm as ‘land next to the M62’. In this situation, file plans are not sufficient for searches to be completed and this results in a delay while another plan is produced.
Using cognitive services for file plan validation means a computer is trained to recognise a polygon shape marked on a map to indicate the area where a search needs to be completed. The technology will ask ‘Is this a good file plan?’ and the solicitor is immediately informed with a percentage of confidence.
The key is trusting the technology. If the confidence level is not 100%, then this can be checked manually and the computer informed of the result. This is how the system learns to be more accurate next time.
There are other potential applications too. You have probably parked your car somewhere which has used number plate recognition to identify your vehicle. Similarly, cognitive services could be used to identify documentation such as passports, confirming the identity of a purchaser.
Computers can be taught to recognise other legal documents too, such as forms for fixtures and fittings or any forms that need address matches.
This offers the potential to cut down on administration, reduce the chance of human error and potentially speed up processes. These developments may sound subtle but, when combined, could transform the efficiency of overall workflows in time.
AI decision-making has further interesting potential applications. Take pricing, for example. During busy periods, the technology could be automatically taught to raise fees, while at quieter times, fees could be reduced to encourage more customer enquiries. Using market-dependent data, AI can make intelligent decisions to influence the bottom line positively.
Consumer expectations and demand are driving the need to speed up the conveyancing process through digitisation. As a result, many forward-thinking law firms are looking into ways to expedite processes as far as possible.
The introduction of new technologies such as cognitive service and AI decision-making could have significant potential to do this. However, technology must be proven and trusted to be introduced safely and successfully over time.
Re: conveyancing
A machine learning system is desperately needed. Legal professionals, time and again show themselves to be at best inefficient, with regards to the above.