LONDON — If you believe the World Economic Forum (WEF), then the world is on the cusp of a “Fourth Industrial Revolution” that will radically change the global jobs market.
Robotics, artificial intelligence (AI), and automation will do away with 5 million net jobs by 2020, according to WEF. Advances in computing mean robots and algorithms will soon be able to do complex tasks once only humans could do.
Blue collar professions such as law, finance, and professional services could feel the same jobs squeeze that white collar industries felt from the rise of robotics in the second half of the 20th century as employers realise more efficiencies through technology.
But today, as we stand at the foothills of this nascent revolution, that’s not quite how it looks. Middle class professions that are using AI and automation aren’t replacing employees with algorithms. Instead, the AI is taking care of all our boring paperwork.
‘It’s very, very quick to learn’
Law firm Slaughter and May have been quick to jump on the AI bandwagon, helping to train startup Luminance’s AI programme to sort through files in the due diligence process. The algorithm crunches through everything from graphs to PDFs and makes them easily searchable and understandable.
Rob Sumroy, a partner at Slaughter and May who heads their fintech practice, told Business Insider that Luminance’s use isn’t so much about driving efficiencies as keeping up with the sheer amount of data in the world.
He told BI: “Whereas in the past there might have been, say, 1,000 documents that needed reviewing in a week, there’s now 10,000 that need reviewing in two days. Inevitably, something’s got to give and what we tend to find ourselves doing is helping the client to take a risk view and deciding which of the documents to focus on in the limited time. You just haven’t got time to go through them all, and there is an inherent risk in that approach.”
Sally Wokes, another partner at Slaughter and May, has helped to train the Luminance AI and estimates it has cut down the time taken for lawyers in the information finding stage of due diligence by at least 50%.
Wokes says: “It’s very, very quick to learn. The algorithm behind this particular software doesn’t rely on word searching. It’s more intelligent than that. For example, the word’s ‘change’ and ‘control’ don’t have to feature in a document for the software to recognise that a clause is, indeed, a change of control clause.”
AI fighting bribery and corruption
Luminance, which is backed by millionaire British software entrepreneur Mike Lynch, is used for due diligence but AI can have all sorts of applications. Bloomberg highlighted recently that JPMorgan has begun using a programme called COIN to do the “mind-numbing job of interpreting commercial-loan agreements,” as they put it.
The Serious Fraud Office (SFO) also used AI in its blockbuster investigation of bribery and corruption at engine maker Rolls-Royce. The SFO used RAVN, a London-based AI startup, to sort through 30 million documents, crunching through up to 600,000 a day.
“The judge in his summing up was great — saying this must be the way forward chaps, to use technology to drive down cost and increase efficiency,” RAVN CEO David Lumsden told BI.
RAVN, founded in 2010 by former Autonomy employees, focused on the legal sector in its early years but Lumsden sees big potential for its programme in the financial sector.
“What a lot of these financial firms do is an awful lot of paperwork,” he says. “They hold reams of paperwork on things they have investigated to see whether it’s worth investing in or not. That information, like in many law firms, just sat there. We let them exploit that information in the knowledge graph way that we do for law firms.”
The beauty of RAVN, and all true AI engines, is that they are a multipurpose tool — they can quite easily be rejigged to ingest totally different information and sort it out.
‘It’s not about replacing people’
Will these programmes one day replace or reduce the numbers of paralegals, junior bankers, and other staff early in their careers?
For the SFO, RAVN did not replace staff but greatly extended their abilities. A team of just 7 people worked on the Rolls-Royce case — crunching through 30 million documents would have taken an age.
Slaughter and May’s Sumroy agrees that AI programmes today are about extending abilities. He says: “It’s certainly not putting lawyers out of work. It means that our lawyers have more time to do what they went through years of legal training for and what they’re good at, which is analysing the output of the review rather than actually doing the review itself.”
He adds: “My first memory of working at Slaughter and May (some 24 years ago) was as a trainee being shipped off to a warehouse somewhere in the north of England. We were helping our client sell a massive chemical business. They had set up a data room in one of the old chemicals warehouses, and all the files that needed reviewing were in there. We did spend days and days and days in there — reviewing contracts.
“Luminance takes the strain. It does the reviewing and enables our lawyers to analyse the outputs and advise on their findings.”
Martin Sweeney, the founder and CEO of Ravelin says: “It’s not about replacing people it’s about giving people the tools to do their jobs better. It’s about making the team more efficient. We’re automating the jobs that people used to do but it’s a lot of the boring stuff and we’re doing it more reliably and consistently.”
Ravelin is a real-time fraud detection tool for online retailers. In the past, a company would set fraud parameters — signs that would flag up a case to be reviewed by a human — but scammers would quickly learn how to circumvent them. Ravelin’s product learns from cases as it goes along, automatically updating. It makes it harder to trick.
As with Luminance, Ravelin’s sales pitch is not about cutting staff but about keeping up with the pace of the world. Ravelin works with “on demand” businesses such as Deliveroo, MyTaxi, and YPlan, all of which must make split second decisions on whether a transaction looks like fraud.
Ravelin flags up suspicious cases for human review and Sweeney thinks the future of AI is humans and algorithms working in harmony. “It’s going to be a combination of humans and machine learning,” he says. “The learnings we have are a direct impact from human input. They tell us when the machine got it wrong.”
Business Insider Emails & Alerts
Site highlights each day to your inbox.