GUEST:
Until recently, when automated technologies emerged there was a gap between the laborers it replaced and the decision-makers who implemented it. The CTO for a car manufacturer could safely implement factory automation that laid off portions of the workforce five steps below him without fearing for his own livelihood. But as AI has matured, it’s begun to climb the corporate ranks, going after positions that require advanced degrees and high IQs.
No longer do the higher-ups make decisions that result in the lower-downs being laid off. Instead, disruptive startups are offering AI services that can replace entire professions with lower prices and more precise results.
The incumbents: A look at recruiting, medicine, and law
Medicine, recruiting, and law are distinct in terms of the education they require and the resulting salaries they earn. But one thing they each have in common is that AI has become sophisticated enough to severely threaten their very existence.
1. Medicine
In 2013, the FDA approved Johnson & Johnsons’s AI anesthesiologist. The machine, named Sedasys, cost only $ 150- $ 200 per procedure (as compared to the $ 2000 it costs for an anesthesiologist to administer). Anesthesiology is one of medicine’s highest paying jobs, the median salary is $ 352,518 per year, and requires four years of training beyond medical school. It’s not particularly surprising, then, that anesthesiologists were staunchly against the newborn machine replacing them. The American Society of Anesthesiologists campaigned aggressively against it until its uses were limited to routine procedures such as colonoscopies. And then in 2016, the machine was discontinued altogether due to poor sales, despite its myriad cost and precision benefits, hospitals were reluctant to replace their highly-valued anesthesiologists.
This is not the only instance of an insurgent machine attempting to replace a high-paying position in medicine. Machine learning algorithms have been able to detect the presence or absence of TB in X-rays with 96 percent accuracy, which is higher than any human radiologist. Researchers with Google trained AI to detect the spread of breast cancer on microscopic images as accurately or better than human pathologists. A recent article published by the Journal of the American College of Radiology (JACR), envisioned a looming future in which AI becomes a routine part of radiologists’ daily lives. And just last week, Alphabet’s health company Verily announced that it developed an AI eye scan that can predict cardiovascular risk factors as accurately as a blood test can. While none of these machines have been widely implemented, and certainly none have replaced humans (yet), the ability for them to do so is certainly there.
2. Recruiting
If Netflix can suggest personalized shows for its users, Uber can match drivers with passengers, and Facebook can personalize its News Feed to users’ interests, then the recruiting industry can use AI to match jobs with candidates. And yet, despite the relative simplicity of this task, recruitment is still a highly manual field, with new hires costing upwards of $ 4,000 over a six week hiring period. What this means is that there’s a massive market for AI-powered recruitment disruptors. Almost every single company in the world is currently wasting money on an inefficient process that results in high turnover and wasted time. In fact, over half of talent acquisition leaders say the hardest part of recruitment is identifying the right candidates from a large applicant pool, which is a task that is extremely easily automated. Companies such as the much-hyped Woo, have hopped onto this solution, using matching algorithms to cut down the time to first candidate profiles to under 24 hours. Notably, the company’s most high-profile customers are all similar disruptors like Uber, Lyft, WeWork, Wix, etc.
Even systems like Woo, though, work in tandem with more manual forms of hiring, such as Lever and Greenhouse, meaning that it does not replace recruiters, so much as streamline one aspect of their jobs. While this is threatening, it’s not so threatening that recruiters would refuse to use it, as anesthesiologists did.
3. Law
AI lawyers have already begun infiltrating the legal profession. From the more mundane LawGeex, which automates contract review, to the truly threatening “Ross” by IBM, which combs through legal documents, offers hypotheses with citations to back them up, and keeps up to date with relevant legal developments, there are myriad robolawyers for hire. And many of them sound a lot like junior associates.
DoNotPay, a chatbot that gives free legal advice for consumer issues with parking tickets, landlords, and retailers, among others, has the explicit purpose of eradicating the legal profession: “The legal industry is more than a 200 billion dollar industry, but I am excited to make the law free,” said the chatbot’s creator. “Some of the biggest law firms can’t be happy!” Another AI-powered law tool was able to predict the verdicts for hundreds of cases heard at the European Court of Human Rights with 79 percent accuracy.
That said, while AI may be able to replace the need for lawyers when it comes to parking tickets, it’s unlikely that it will eradicate the profession, considering that interpretations of the law change over time. And without human training data, a machine would not be able to adapt. Furthermore, machines trained through precedent will invariably hold onto human biases, such as racism and sexism. In fact, in 2016 multiple news outlets revealed that a software that predicted the likelihood of repeat offenses was mistakenly flagging black defendants at almost twice the rate as white ones.
Why disruptors haven’t gotten market penetration
Those with the most power to adopt automation are also those with the most to lose by adopting it. Despite a machine’s ability to administer anesthesia with greater precision than a human, no anesthesiologist would ever encourage the use of this machine, especially considering that many doctors have extremely burdensome student loans.
This is indicative of a larger endemic problem with automation. As AI becomes more adroit at taking highly skilled positions, we run up against a crisis point: Those who can choose to use it would essentially be sealing their own obsolescence. Processes within these professions that save time on menial tasks will be successful; those that position themselves as robo versions of the profession will likely not.
In the end, whether or not a company succeeds at infiltrating highly skilled jobs with automation will depend upon messaging. The lawyer chatbot, DoNotPay, was successful only because it went straight to consumers, if it had used the same messaging with firms, it would likely not have been implemented. If more companies choose to follow this route, direct-to-consumer services, there will need to be policy changes to protect these professions. After all, a world without doctors, lawyers, or recruiters would be lacking far more than just these professions, we would see a ripple effect to graduate programs, professorships, and employment rates in general. And perhaps we should have considered this sooner. As AI becomes increasingly sophisticated, we will either need to curb its applications via policy, or we will have to completely overhaul our income and employment infrastructures.
Sascha Eder is the cofounder and COO of NewtonX, a general program for molecular dynamics simulations beyond the Born-Oppenheimer approximation.
