My latest (draft!) paper, When AIs Outperform Doctors: The dangers of a tort-induced over-reliance on machine learning and what (not) to do about it is, I hope, special. I had the good fortune to co-author it with Canadian polymath Ian Kerr, and with Joëlle Pineau, one of the world’s leading machine learning experts.
Here’s the abstract:
Someday, perhaps soon, diagnostics generated by machine learning (ML) will have demonstrably better success rates than those generated by human doctors. What will the dominance of ML diagnostics mean for medical malpractice law, for the future of medical service provision, for the demand for certain kinds of doctors, and—in the longer run—for the quality of medical diagnostics itself?
This article argues that once ML diagnosticians, such as those based on neural networks, are shown to be superior, existing medical malpractice law will require superior ML-generated medical diagnostics as the standard of care in clinical settings. Further, unless implemented carefully, a physician’s duty to use ML systems in medical diagnostics could, paradoxically, undermine the very safety standard that malpractice law set out to achieve. In time, effective machine learning could create overwhelming legal and ethical pressure to delegate the diagnostic process to the machine. Ultimately, a similar dynamic might extend to treatment also. If we reach the point where the bulk of clinical outcomes collected in databases are ML-generated diagnoses, this may result in future decision scenarios that are not easily audited or understood by human doctors. Given the well-documented fact that treatment strategies are often not as effective when deployed in real clinical practice compared to preliminary evaluation, the lack of transparency introduced by the ML algorithms could lead to a decrease in quality of care. The article describes salient technical aspects of this scenario particularly as it relates to diagnosis and canvasses various possible technical and legal solutions that would allow us to avoid these unintended consequences of medical malpractice law. Ultimately, we suggest there is a strong case for altering existing medical liability rules in order to avoid a machine-only diagnostic regime. We argue that the appropriate revision to the standard of care requires the maintenance of meaningful participation by physicians in the loop.
I hope that it will be of interest to to lawyers, doctors, computer scientists, and a range of medical service providers and policy-makers. Comments welcome!
Recently I’ve been posting pretty much nothing but videos. There two reasons. First, to the limited extent I’ve had spare time, I’ve been spending it reading political news with more horror than inspiration. Second, there hasn’t been much spare time: I just sent in a revised first draft of an article that will in due course be published by the Arizona Law Review. Meanwhile I also did a first draft of a conference proposal on some AI issues to send to a co-author. And, last week I discovered that the other conference proposal, on a privacy issue, that I thought was due at the end of this week turns out to be in fact a call for finished papers. Fortunately, I’d already started, but getting to the finish line is taking most of my time at present and seems likely to do so through the coming weekend. After that, another large privacy-related project looms in the distance. Meanwhile, of course, Jotwell is humming along. We recently started a new Contracts section.
I just had to turn down an invitation to speak on AI and robotics in Hong Kong next month because I couldn’t find enough times for all the makeup classes on such short notice. That hurt, as I’d have loved to go. But if you were maybe thinking of asking me to come talk about the exciting things I’m working on, some of which I’ll blog about when they are a little more baked, please be aware I’m on writing leave next semester, and I will have much more flexibility than usual–and, I hope, more work product too.
Just modified my writing tips for students to add “A Few Words About File Names”:
Most of our students will spend at least some part of their careers working in large organizations where they share electronic documents. Students should get in the habit of giving their documents good file names. Thus “paper draft” is always bad — imagine if everyone calls their documents that!
Some organizations have naming conventions, but if not, be sure the file name has three things: your surname (if the primary author, so you get the credit), something substantive (not “paper” or “motion” but “Jones-dismiss-motion” or whatever), and a version number. You usually do not need a date, the software will do that automatically. For papers it’s probably a good idea to put your surname first, then a summary form of the paper title, as that way over the course of time all your submissions to your professor will be grouped together and easy to find.
It is good a habit to avoid spaces in file names (use – or _) so that files translate well to UNIX systems if they ever touch one. Do not use any other odd non-alphanumeric characters as it can mess up some older file systems.
All fairly obvious, but you would be amazed how many files I get from students called “paper”.
Robot Law is the front of their homepage today, and the subject of an article, Rock ‘em, sock ‘em, cross examine ‘em.
This is exciting: just got my first copy of “Robot Law,” a book I edited with Ryan Calo and Ian Kerr. I suppose I might be a little biased, but I think it’s a pretty darn good collection that will give anyone interested in how society will cope with robots plenty to think about.
Robot Law is apparently going to list for $165 when it’s out in (very) late March, which is a lot, but you can pre-order it for less, or buy an online copy for much less. Meanwhile, however, you can peek inside, and read my introductory essay which gives you a tour of the wonderful contributions by our extraordinarily varied contributors. This is not a book just by some law profs: it’s an attempt to do real interdisciplinary work and, more importantly, to foster an ongoing series of interdisciplinary conversations.
Of course, the real-life place where we do that is at We Robot — registration for this year’s conference is now open and the early-bird discounted registration ends Friday.