- It is not ‘can a machine ever exercise judgement?’ The right question is, ‘To what problem is judgement the solution?
- What machines are incredibly good at doing is processing far larger volumes of data than human beings and running algorithms and routines through them
- In a world where these systems and machines become more capable the most challenging question will become one of distribution
For centuries doctors, accountants, lawyers, teachers and architects have solved problems too complex for most of us. But today, the future of these professions are seriously in question as artificial intelligence, non-thinking machines and automated systems are poised to do the jobs of human experts at a fraction of the cost.
Such a prediction triggers one of two responses, says Daniel Susskind, co-author of The Future of the Professions. Young professionals are often furious; consumers are quietly optimistic.
“For consumers it is very exciting. There will be far more affordable access to expertise that has traditionally been locked up in the heads of the professionals,” he says.
Or put another way: “The professions have created a Rolls Royce service and most people are walking.”
Susskind wrote the book with his father Richard Susskind, a lawyer. It is based on 100 interviews across eight professions, from law through to accounting, consulting and the clergy.
A barrage of statistics sets the scene. In 2014, 48 million Americans used online tax preparation software rather than traditional tax accountants. The best-known legal firm in the US is no longer a traditional law firm, rather the legal advice platform legalzoom.com
One architectural firm recently used flying robots to assemble a building; doctors talk to patients via Skype; even the Catholic Church has an app that tracks sins as it prepares its flock for confession.
So is this the start of a social revolution?
“We see two futures,” says Susskind.
“The first is incremental transformation. This is a more efficient version of what we have today, not a big bang revolution. The second future is long term. Here, technology actively displaces the work of traditional professions. Over time more of the tasks done by accountants are going to be done either by different types of people or ‘increasingly capable systems’ or machines, or by the two.”
There are two major factors that are shaping this future. Firstly, the failure of “the grand bargain”. This is how the professions traditionally provide services to the exclusion of others, instead of as gatekeepers of knowledge. Not only are their processes expensive and antiquated in our internet-based society, it’s as if there is “an intentional obfuscation” of the work they do, according to Susskind.
Secondly, to quote Patrick Winston, one of the fathers of artificial intelligence (AI), there are lots of ways of being smart that aren’t smart like us. This idea is central to the book.
Other ways to be smart
“We tend to think that the only ways of performing tasks like human beings is to replicate the way human beings perform,” explains Susskind.
“A professional will say, ‘What I do in my job is exercise judgement and as a machine can’t think or reason, it can never exercise judgement, so these particular tasks are safe from automation.’”
This is the wrong question, he says.
“It is not ‘can a machine ever exercise judgement?’ The right question is, ‘To what problem is judgement the solution? Why do we call upon human experts in this particular case to exercise judgment?’ The answer is uncertainty,” he says.
“So a better question is, ‘Can a machine deal with uncertainty better than a human being?’ The answer to that is of course they can. What machines are incredibly good at doing is processing far larger volumes of data than human beings and running algorithms and routines through them.”
Famously, the first evidence of this was back in 1997 when IBM’s Deep Blue beat world chess champion Garry Kasparov. The computer was able to calculate up to 330 million moves per second whereas Kasparov — at best — could consider about 100 moves per go. The computer won through brute force processing power. Fourteen years later the supercomputer, IBM Watson, stunned Jeopardy audiences when it won hands down against two of the American quiz show contestants.
Today’s second wave of AI machines are no longer being built to replicate human reason. It is this that the Susskinds define as the “artificial intelligence fallacy”.
In the second wave of AI, many of these systems and machines are outperforming human beings by working in fundamentally different ways, Susskind says.
“They are not like us. So much time is spent thinking about whether AI systems are conscious or thinking. That is not the most important question from the point of view of the professions.
“What is most important is to ask if these machines can outperform human beings even if the way they do it looks very different to the way human beings do it? The answer to that is increasingly ‘yes’.”
In America, Daniel Katz, a law professor has created a machine that successfully predicts the outcome of US Supreme court decisions. This system knows nothing about the law, it relies on data from past Supreme Court cases.
“This lesson isn’t just true for judgement but across all faculties of human beings,” continues Susskind.
“It’s a mistake when people say a machine can never exercise creativity or empathy.”
It’s a mistake when people say a machine can never exercise creativity or empathy.
Future roles for accountants
If you’re seeing visions of Stanley Kubrick’s HAL 9000 or your inner sceptic is doing cartwheels, it’s worth knowing that Richard Susskind has spent the past 40 years writing about how technology is going to affect the legal profession.
In 1996 when Richard published The Future of Law, predicting that the dominant way lawyers and their clients would communicate would be through email, the Law Society of UK and Wales declared that he shouldn’t be allowed to speak in public and he was bringing the profession into disrepute.
Since then, Susskind senior has written a clutch of heavy-weight legal tomes and when The Future of Professions was published in the UK in October 2015, Acuity heard that lawyers across London rushed out and bought a copy.
“I heard accountants did the same,” says his son.
“The accounting profession, probably perhaps more so than the law, has adopted and taken up the book. There’s been a huge amount of interest, both from practitioners and professional associations as they ask themselves what their role might be in the twenty-first century. In particular how associations can help their members understand the changes that are taking place and give strategies to deal with it.”
Disappointingly, it is the medical profession that is ostrich-like.
The Susskinds’ joint collaboration brings an inter-generational perspective to their research. To write it, father and son initially holed up in a “bunker” in the city of London, brought in white boards, post-it notes, hundreds of their books, and trays of smoked salmon sandwiches as “mind fuel” to thrash out their argument.
Over the four-year period Susskind junior says: “We agreed to argue until we agreed on every point.”
Intriguingly, he describes 2014/15 as a watershed.
“I think people wouldn’t have been so receptive to the book if it had come out earlier. Something happened during this time — people’s attitudes to technology and what it was capable of doing, companies like IBM working with Watson, Google’s work on AI, other books such as The Rise of the Robots by Martine Ford. All of them said something significant is changing.”
Susskind is both optimistic and pessimistic. In the medium term, the scenario is less about unemployment as about redeployment. The book describes a new class of “para-professionals” who will be equipped to undertake routine tasks previously done by senior professionals. It identifies seven models for the production and distribution of practical expertise and 12 new roles, which range from “empathisers” who can provide reassurance and empathy, to data scientists and systems engineers.
The longer term, he admits, is more troubling.
“It’s hard to avoid the judgement that there will be a decline in demand for human professionals. In a world where these systems and machines become more capable the most challenging question will become one of distribution. It will be difficult to live in a society where some very productive people work while less productive people don’t work. The problems will be distributional and political and how we reconcile those things.”
For all of us, some serious food for thought.
This article was first published in the December 2016 issue of Acuity magazine.