I completely disagree with the article's assertion that 'knowledge professions like ... lawyers' will be relatively unaffected. 'DoNotPay' (the robot lawyer) has already helped overturn millions of dollars of parking fines. NPR article on the subject: https://www.npr.org/2017/01/16/510096767/robot-lawyer-makes-...
The thing about lawyers is that they're often able to create work for themselves. They can lobby for laws that create opportunity if automation affects them too much. They can create class actions out of nothing and recruit their clients. They're also the ones that write the laws, and you can always expect that the rules of the "game" will always favor those with the most input into those rules.
Except the real trend has been the exact opposite. We're now in the fourth decade of lawyers (legislators and judges) narrowing the laws you're talking about: class actions, discrimination claims, environmental claims, and securities claims. As a result, the legal sector as a percentage of the economy has been shrinking since the late 1980s: https://lawschooltuitionbubble.files.wordpress.com/2012/12/r....
I personally think that's a good thing because the economy was over-regulated in the 1960s and 1970s. But in some areas, it's arguably gone too far. Judges and legislators are so self-concscious about excessive litigation that they've significantly hampered the ability of claimants to recover when they should be able to do so. E.g.Wal-Mart v. Dukes.
Agreed, this goes for other professions as well such as accountants and real estate agents. There always seems to be some new complexity that requires the profession. That being said I do think it will put a dent in the number of professionals. Sites like homie.com for real estate already seem to he effecting real estate personnel.
Some of these automation companies are doubtless being run by people with legal backgrounds. I can't imagine all lawyers are interested in preserving the status quo.
I happen to be watching tree trimmers work on a tree from my office window, so I entered that into their form. Very high automation risk (77%)... but I'm looking at this work, and it's absurd. Automating it would be extremely difficult, and serve no purpose since there's nothing particularly expensive or wasteful about the current process. To be fair, it identifies plumbers as having a fairly low automation risk, which I think is reasonable.
Still, I'm skeptical how they identify automation risk, and I suspect they get it wrong in a way that makes small cities seem more vulnerable.
okay, but what if there is a home robot with significant dexterity for generic tasks exists for other reasons. would you stop its owner from using that for tree trimming?
my rule of thumb is simple, if a job only exists because of some amount of cognition needed with fairly simple decision making it will be gone regardless of the low pay. also, some higher end jobs like investment management are already getting automated given that they follow the same meme. its just that their cognition/pattern recognition tasks involved sifting through balance sheets (e.g. NYSE:AIEQ).
Investment management? Sure, that seems pretty easy to automate.
A home robot that can trim trees is FAR more complex. It needs to know about trees and what to trim, and talk to the customer. Maybe doable. It needs to analyze the environment to look for any dangers, fragile things, property lines. It needs to navigate heavy equipment. It needs to make a series of cuts, and make sure nothing too large falls after cutting. It needs to collect and dispose of the waste. This is way beyond any robotics we're seeing.
In a sci-fi world where robots can do EVERYTHING, they can also trim trees.
But we are not just talking about replacing humans entirely. How much can more advanced robot-tools speed up the process? Like the human operator talks to the customer, makes some gestures in AR to signal where to cut while improved robots / tools take care of lengthy / tedious work.
I totally see this happening and requiring much less human labor. And as with most automation, it will be a gradual process and not a one step replacement.
robot dexterity and cognition are both very very hard problems. Once we solve that, we just don't get human hand equivalents. We can make very tiny hands that can manipulate cells and perform intricate surgeries, or huge hands, that can move boulders and build a house in a day, or hands with 100s of snake like fingers manipulating 100s of things simultaneously and running a click farm.
Once we can build artificial brains more powerful than humans, and artificial muscles more dexterous than humans, almost every profession is at risk. In addition if machines are not made in a factory, but replicated organically like seeds because we figured out the magic DNA, then we've essentially created von-neumann probes and can conquer the rest of the galaxy.
Its possible that some mundane examples will be automated, but the law is not a deterministic code that can simply be interpreted and executed. Automation may make the job less tedious and might reduce the numbers of lawyers needed (or more likely, paralegals).
Many entry-level lawyers have historically been involved in document discovery. AI-based software and having a significant impact on the job market for freshly minted lawyers. Software now performs ediscovery.
From first hand experience, I would say that the willingness of clients to outsource discovery to cheap contract lawyers has had a far greater impact on the job market for fresh graduates than automation of discovery. The software for electronic discovery just isn't very good or smart. It also requires a ton of up-front work to train the model which makes it not cost effective except for the largest document reviews.
I've heard this example bandied about every time this topic comes up, but does this actually mean that those entry-level lawyers would be out of a job? Or does it simply mean that they would be freed up to perform other tasks? Wouldn't that be a win for a law firm that's presumably trying to grow and take on more clients?
The current job market for lawyers is kind of messed up. When in law school you can be applying and accepting job offers clerking or working at law firms 2 years before the start date.
I don't know what even more increased pressure on that job market would do to it.
You accept an offer conditional on passing the bar. But even once you're out of school, if you want to do a clerkship you have to apply for it 1-2 years in advance.
Quite a few paralegal jobs have been removed simply by adding document search capabilities. Automation that cuts the work down by 50% can have huge impacts even without dealing with all the edge cases.
I was under the impression that DoNotPay worked well because text is fairly boilerplate (if so can you really call that automation?). Plus, I don't think many people pay lawyers for parking tickets.
As far as law in general, I have heard convincing arguments that it would be difficult to make ML for defense and judging, given the tainted historical data (and likely that historical data will always be tainted as our morality and laws change).
Also, from my (very much outsider) observation of the legal profession there seems to be a lot of boring work that probably could be automated. So maybe not lawyers themselves, but legal helper workers, will increasingly be replaced.
It seems to me that knowledge worker support staff (paralegals, PACs, etc) are the current low hanging fruit in terms of automation. Firms without them will be able to turn higher profit on less revenue.
Right, knowledge professions are actually innately geared towards being automated. However its worth noting that the higher levels of these jobs cannot be automated, as well as the human aspect of the courtroom (judges). Whatever algorithm you apply is going to need datasets (i.e) parking ticket cases so it cant operate on the higher end of the legal system. The judicial system also provides a place for the meaning of laws to be decided by humans, you cant automate that.
The legal services industry in the US is around a $250 Billion industry, example of a automation of 0.004% of that doesn’t mean anything. Again the Toyota story doesn’t seem to contain any long term, industry wide statistical analysis. Without that it’s just an anecdote.
Also in reference to manufacturing processes, it appears that Toyota has learned some hard lessons in automation and is actually moving in the opposite direction: http://www.thetruthaboutcars.com/2011/02/toyota%E2%80%99s-se...