Much of the debate around the new AI-driven phase of automation centres on the likelihood of job losses once software replaces humans performing repetitive tasks involving large quantities of data processing.
Yet relatively little attention has been given to automation’s potential for social good, perhaps because innovators naturally gravitate towards areas that will be most lucrative. So it is that we have been blessed with robo-advisers and finance chatbots, whose impartiality is called into question by the motives of their paymasters.
But then there’s Joshua Browder, an automation vigilante dubbed the ‘Robin Hood of the internet’. He taught himself to code at the age of 12, and six years later in 2015 he built DoNotPay, a website that could automatically appeal users’ parking tickets – for free.
“It became obvious when I was writing [appeals to local authorities] that I was writing the same letters over and over again,” says Crowder. “I thought this is something that can be repeated and automated. I used the Freedom of Information Act to look for the top 12 reasons parking tickets are dismissed. Then I got lawyers and used available data online to generate generic template letters for each of these reasons, which I then put into this chatbot that gets the details from you to generate a letter which is automatically sent to the local authority.”
Since then, over 175,000 users have successfully appealed their parking tickets, saving motorists in London and New York an estimated $5m in parking fines.
Now, Browder has expanded his robo-lawyer to cover several more areas of the law using his chatbot model. Having started out with a system that diverted funds from local authorities, he developed a new tool with homeless charity Centrepoint that would save governments money while helping people with more drastic problems than a parking fine.
“We have this broken system where instead of housing the homeless directly, the government will pay a lawyer huge sums of money to file an application that is then sent back to the government. My chatbot will talk to a person to identify if they’re eligible for government housing and – as with the parking tickets – file an application which is sent to the government.”
Again, the service is obviously free, and Browder has made his code available to the public to the extent of launching a platform whereby anyone with knowledge of a specific legal issue can create a law-bot using a drag-and-drop builder. Nevertheless, he recognises the need to make the platform relatively limited to avoid misuse, for example through facilitating tax evasion.
RAVN has built a robot called ACE that organises, discovers and summarises data with an essentially human level of understanding. ACE lightened the seven-strong SFO investigation team’s load of 30m documents by sorting through up to 600,000 on a daily basis, separating them into ‘privileged’ and ‘non-privileged’ information without the threat of human error. Carried out by humans alone, the process would have taken months.
Now, the SFO is considering using the software in future investigations, hoping that ACE’s ability to repetitively sift through vast stacks of legal documents will free lawyers up to focus on more valuable work.
What is striking about automation is how surprising and varied its applications can be. And – particularly in Joshua Browder’s case – we can see that its capacity for social good is vast, albeit dependent on who develops and controls the technology.