Startups spy an opportunity in the power of AI and automation

Have you heard the one about the Chinese mobile phone factory that made one change and increased productivity by 250 per cent?

In news that won’t have been very funny to employees, the Changying Precision Technology replaced 90 per cent of its human workforce with machines.

The workforce at the factory in Dongguan City shrank from 650 people to just 60 after the company invested in 60 robot arms that can work 24 hours a day across 10 production lines. Luo Weiqiang, general manager of the factory, says the number of employees could be reduced to just 20.

Factory jobs are the most obvious casualty in the march towards automation. Almost half of all the manufacturing jobs in the UK could soon be done by machines. Transport and retail jobs will also be transformed in ways that are not beyond the imagination for anyone who has used a self-checkout at a supermarket or been following Tesla’s experiments with driverless cars.

Beyond the 30 per cent of UK jobs that are expected to be replaced by machines by 2030, artificial intelligence is creeping into other less likely industries. Professions once considered essentially human are employing robots to complement human judgement, opening up opportunities for startups.

Euan Cameron, leader of artificial intelligence at PwC UK, says computers are good at jobs that are composed of different tasks. “They are good at locating knowledge, recognising patterns, understanding natural language and continuous learning,” he says.

They are less good at jobs with characteristics that are seen as uniquely human like common sense, morality, creativity, ethics and emotional intelligence. “Most jobs are a combination of those. So the extent to which a job will be disrupted depends,” Cameron says.

Industries including teaching, law and medicine have become a battleground for startups and investors. They are eager to disrupt first with technology like WriteLab, an online writing coach, or Redox, which helps organisations share healthcare data.

Luminance is a start-up that using computers to process large data sets to help lawyers and law firms do due diligence on contracts. At the moment, contracts are reviewed by newly-qualified lawyers who spend long hours over late nights and weekends reading hundreds of pages, a process which can be complicated by tiredness and errors.

Emily Foges, Luminance's chief executive, says the company allows junior and trainee lawyers get back to more interesting work analysing the data as they did before contracts were drawn up digitally, producing hundreds, possibly thousands of digital documents.

“Trying to find clauses is like going through a needle in a haystack,” Foges says. “The process of due diligence is no longer fit for purpose. The idea that you can read everything about a contract and a company in a few weeks is no longer possible.”

Computers, however, are excellent at spotting anomalies in data sets more quickly than humans can. It might identify a missing clause in a contract that can then be added in by the lawyer. Luminance says this process doesn’t replace the role of lawyers but helps them to categorise, review and analyst the documents, improving efficiency by at least 50 per cent.

Artificial intelligence can also help ordinary people to do work that once required legal training. DoNotPay is an app that was created by Joshua Bowder, a 20-year-old Stanford University student from London. It acts as a paralegal service, helping drivers dispute parking tickets by giving everyone the same legal access. The online bot has helped overturn more than 200,000 parking tickets in London, New York and Seattle.

It is harder to convince people of the case for artificial intelligence in professions like teaching and medicine, which require higher levels of emotional intelligence.

Cameron says one problem is the ambiguity of the term. Artificial intelligence is essentially about using machines to do things normally associated with human capabilities. “That almost always has a learning element to it, whether that involves learning data, insights or judgement. Judgement tends to be what people are nervous about delegating,” he says.

Teachers, for example, are valued for their pastoral role with children, comforting and guiding as they grow.

AI could cut the hours teachers spend on paperwork, planning and assessment or transform the way lessons are delivered, so the child does the bulk of the work at home and exercises with the teacher in the classroom.

“AI devices can monitor the behaviour of the child, such as which answers they are getting right and wrong, which questions they are selecting. Then you have this data for what kind of interventions are needed in terms of additional modules,” Cameron says.

There are privacy concerns about collecting data this way. A pilot by a company called InBloom in New York in 2014 collapsed after it ran into issues.

Confidentiality is an even bigger issue in healthcare. As is trust: patients are likely to be unsure about the idea of robots conducting surgery for now. “In medicine you need a bedside manner,” Cameron says.

“But behind the scenes clinicians are performing diagnosis where the assistance of an AI agent can improve the quality and efficiency.”

A computer’s ability to spot anomalies in data can help clinicians identify unusual signs and symptoms in patients better. “A clinician can do a pretty good jobs and a trained computer can do a good job but together they can do an even better job,” Cameron adds.

Ursula Huws, professor of labour and globalisation at the University of Hertfordshire, says that while it is true that artificial intelligence will replace some jobs, it will also create jobs, like those at startups developing the technology. “As fast as you have jobs disappearing because of automation, you have new jobs to make the robots, clean the robots, deliver them, mine the raw materials,” Huws says.

Cameron agrees that the number of jobs in the marketplace will stay broadly flat as artificial intelligence develops, but adds that the technology should provide a boost to the economy by increasing productivity. “The overall economic impact will be beneficial,” he says.