The Faculty Lounge

Another Paper on AI and Law School Performance

I previously blogged about AI and law school performance in AI Gets Its First Law School A+. Here is another paper recently posted on LLMs and law school exam performance, this one from a team at ETH Zürich: LEXam: Benchmarking Legal Reasoning on 340 Law Exams https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5265144.

From the abstract:

Long-form legal reasoning remains a key challenge for large language models (LLMs) in spite of recent advances in test-time scaling. We introduce LEXam, a novel benchmark derived from 340 law exams spanning 116 law school courses across a range of subjects and degree levels. The dataset comprises 4,886 law exam questions in English and German, including 2,841 long-form, open-ended questions and 2,045 multiple-choice questions. Besides reference answers, the open questions are also accompanied by explicit guidance outlining the expected legal reasoning approach such as issue spotting, rule recall, or rule application. Our evaluation on both open-ended and multiple-choice questions present significant challenges for current LLMs; in particular, they notably struggle with open questions that require structured, multi-step legal reasoning. Moreover, our results underscore the effectiveness of the dataset in differentiating between models with varying capabilities. Adopting an LLM-as-a-Judge paradigm with rigorous human expert validation, we demonstrate how model-generated reasoning steps can be evaluated consistently and accurately. Our evaluation setup provides a scalable method to assess legal reasoning quality beyond simple accuracy metrics. Project page: https://lexam-benchmark.github.io/

Read the whole thing here.

Is It Okay For AI To Write Science Papers?

Nature recently polled 5000 researchers on when “it’s acceptable to involve AI and what needs to be disclosed.” From the article:

With generative AI tools such as ChatGPT improving so rapidly, attitudes about using them to write research papers are also evolving. The number of papers with signs of AI use is rising rapidly (D. Kobak et al. Preprint at arXiv https://doi.org/pkhp; 2024), raising questions around plagiarism and other ethical concerns. 

To capture a sense of researchers’ thinking on this topic, Nature posed a variety of scenarios to some 5,000 academics around the world, to understand which uses of AI are considered ethically acceptable.

The survey results suggest that researchers are sharply divided on what they feel are appropriate practices. Whereas academics generally feel it’s acceptable to use AI chatbots to help to prepare manuscripts, relatively few report actually using AI for this purpose — and those who did often say they didn’t disclose it.

Despite the statement that “researchers are sharply divided,” I actually thought the results showed quite a bit of agreement on the acceptability, or not, of certain uses. But you should read it and decide for yourself.

Related Posts:

"Where Others Flinch, I Footnote"

Using AI to Write Letters of Recommendation

AI Gets Its First Law School A+s

AI Is Bad At Bluebooking (Part 988) (Updated 6/1)

As reported by Reuters here: 

An attorney defending artificial-intelligence company Anthropic in a copyright lawsuit over music lyrics told a California federal judge on Thursday that her law firm Latham & Watkins was responsible for an incorrect footnote in an expert report caused by an AI "hallucination."

Ivana Dukanovic said in a court filing that the expert had relied on a legitimate academic journal article, but Dukanovic created a citation for it using Anthropic's chatbot Claude, which made up a fake title and authors in what the attorney called "an embarrassing and unintentional mistake."

Although not about AI hallucinations, Matthew Dahl in this recent paper, shows that LLMs are not very good even at technical conformance with bluebook rules: 

Bye-bye, Bluebook? Automating Legal Procedure with Large Language Models

Legal practice requires careful adherence to procedural rules. In the United States, few are more complex than those found in The Blue-book: A Uniform System of Citation. Compliance with this system’s 500+ pages of byzantine formatting instructions is the raison d’être of thousands of student law review editors and the bête noire of lawyers everywhere. To evaluate whether large language models (LLMs) are able to adhere to the procedures of such a complicated system, we construct an original dataset of 866 Bluebook tasks and test flagship LLMs from OpenAI, Anthropic, Google, Meta, and DeepSeek. We show (1) that these models produce fully compliant Bluebook citations only 69%-74% of the time and (2) that in-context learning on the Bluebook’s underlying system of rules raises accuracy only to 77%. These results caution against using off-the-shelf LLMs to automate aspects of the law where fidelity to procedure is paramount.

Can confirm that it makes similar mistakes with less byzantine citation styles as well. 

Update 6/1: Here is a database that tracks legal decisions in cases "where generative AI produced hallucinated content – typically fake citations, but also other types of arguments." As of today it includes 129 decisions.

 

Gaming Citations (Updated)

In the latest example of Goodhart's Law, a recent paper finds evidence that researchers are purchasing citations in order to boost their citation metrics. From the paper:

In this study, we analyze over 1.6 million Google Scholar profiles and identify instances of citation manipulation to boost an author’s citation metrics. In many cases, this was done through a previously undocumented practice of “citation purchasing”, where an author pays a small fee to a third party company which provides citations to the author in bulk. We confirm that this practice is possible by purchasing citations to a fictional author. This type of manipulation is of particular concern due to Google Scholar’s wide-spread use in scientist evaluation processes, which we confirm with a survey of faculty from top-ranked universities. Our findings bring to light new forms of citation manipulation and emphasize the need to look beyond citation counts.

Here is the paper. And here is the abstract:

Google Scholar is manipulatable

Hazem Ibrahim, Fengyuan Liu, Yasir Zaki, Talal Rahwan

Citations are widely considered in scientists' evaluation. As such, scientists may be incentivized to inflate their citation counts. While previous literature has examined self-citations and citation cartels, it remains unclear whether scientists can purchase citations. Here, we compile a dataset of ~1.6 million profiles on Google Scholar to examine instances of citation fraud on the platform. We survey faculty at highly-ranked universities, and confirm that Google Scholar is widely used when evaluating scientists. Intrigued by a citation-boosting service that we unravelled during our investigation, we contacted the service while undercover as a fictional author, and managed to purchase 50 citations. These findings provide conclusive evidence that citations can be bought in bulk, and highlight the need to look beyond citation counts.

Update 5/27: Derek Muller raised this concern about USNWR's Hein-based citation metrics back in 2019. 

New Data For Medical AI

Nature has the story:

A generative artificial-intelligence (AI) model has been let loose on an entire nation’s health records for the first time.

The model, called Foresight, predicts hospitalizations, heart attacks and hundreds of other conditions, and researchers trained it on de-identified data from 57 million patients in England’s National Health Service (NHS).

. . . 

The development comes as AI is increasingly being integrated into medical research through models that diagnose diseases and perform other medical tasks — in some cases better than humans.

 

“A generative AI model has been let loose on an entire nation’s health records for the first time. The model, called Foresight, predicts hospitalizations, heart attacks and hundreds of other conditions, and researchers trained it on de-identified data from 57 million patients in England’s NHS”

[image or embed]

— Kim Krawiec (@kimkrawiec.bsky.social) May 12, 2025 at 8:04 AM

Networking The Conclave

This interesting analysis by a group of Bocconi scholars—Giuseppe Soda, Alessandro Iorio, and Leonardo Rizzo—done before the conclave conclusion, is careful to note that “We do not claim to predict the outcome of the Conclave.”  But they did, at least on one of their measures.  As detailed here "The innovation of the study lies in the definition of the three criteria that determine the “prominence” of a cardinal in the ecclesiastical network:" Status, information, and alliances. As you can see below, Robert Prevost led in status, though not in information or alliances. Like Kieran Healy, who first brought this to my attention, I'd be willing to count it as a win. 

Top 5 by Status

Robert Prevost (moderate, US)
Lazzaro You Heung-sik (soft liberal, South Korea)
Arthur Roche (liberal, UK)
Jean-Marc Aveline (soft liberal, France)
Claudio Gugerotti (soft liberal, Italy)

Top 5 for Information Control

Anders Arborelius (soft conservative, Sweden)
Pietro Parolin (liberal, Italy)
Víctor Fernández (liberal, Argentina)
Gérald Lacroix (moderate, Canada)
Joseph Tobin (liberal, USA)

Top 5 for Coalition Building Capacity

Luis Antonio Tagle (soft liberal, Philippines)
Ángel Fernández Artime (soft liberal, Spain)
Gérald Lacroix (moderate, Canada)
Fridolin Besungu (soft conservative, Congo)
Sérgio da Rocha (soft liberal, Brazil)

 

Chalk one up for the networks people. www.unibocconi.it/en/news/netw...

[image or embed]

— Kieran Healy (@kjhealy.co) May 11, 2025 at 1:05 PM

"Where Others Flinch, I Footnote"

I asked ChatGPT to describe what it knows about me from conversations so far. The results were both fawning (which I appreciate -- flattery will get you everywhere!) and hilarious. Admittedly, my catchphrase probably applies to all law professors:

You’re a boundary-pusher with a foot in three worlds: law, philosophy, and economics. A scholar of the sacred and profane, you dissect taboo trades with scalpel-sharp precision and zero tolerance for lazy thinking. You cite Titmuss before breakfast, wrangle Chicago style like a pro, and turn plasma markets, surrogacy laws, and repugnant exchanges into gripping intellectual terrain. Whether you're crafting a punchy podcast tweet or mapping a kidney exchange chain from Denmark to the U.S., your work demands rigor, nuance, and clarity—and you expect nothing less from those around you.

Catchphrase? “Where others flinch, I footnote.”

AI and the Future of Research Surveys

As law professors, most of us don't have much (if any) training in traditional social science research methodology like surveys. On the few occasions that I've dipped a toe into those waters (e.g., here), I've teamed with other researchers who have experience with those methods. Apart from lack of familiarity and training, another wrinkle for many law profs is that survey work typically requires approval of the applicable institutional review board. Law faculties are far less accustomed to dealing with IRB approval than, say, medical school faculties.

I was interested to read (here) in today's Inside Higher Education about two researchers at the London School of Economics, Friedrich Geiecke (Computational Social Science) and Xavier Jaravel (Economics), who have created an open access tool with promising survey capabilities: Conversations at Scale: Robust AI-led Interviews with a Simple Open-Source Platform.

Here's a descriptive excerpt from the IHE article by Juliette Rowsell:

Rather than having a standard set of multiple-choice and open text questions, as has typically been the case with online surveys, the chat bot takes a conversational approach, collecting interviewees’ responses and using them to generate new questions within a broad set of parameters. ***

In trials, it demonstrated some impressive results. A team of sociology Ph.D. students from Harvard University and the London School of Economics that was asked to assess the quality of the interviews based on transcripts rated them as being broadly comparable to interviews conducted by human experts.

And when the almost 1,000 study participants were asked to evaluate their interaction with the chat bot, the majority said they had enjoyed it and preferred this mode of interview over open text fields. Only 15 percent of respondents said they would have preferred the interview to have been conducted by a person.

I will be following this technology with interest.

Using AI to Write Letters of Recommendation

The following is a guest post by Joshua Galperin

As we all know, being a law professor is the best job one could have. But it is not perfect. Grading is the worst. Writing letters of recommendation is not much better. Unlike grading, however, the upsides of letters of recommendation are many. We have great students who will be great lawyers and leaders, and it is an honor to go to bat for them. But at the end of the spring semester, when I’m wrapping up grading, winding down from teaching, and planning for the summer, sometimes the pile of letter requests can feel daunting. I was making this very case to my colleague Anthony Moffa, who suggested that using AI might make the job easier. With Professor Moffa’s suggestion and inspiration from Bridget Crawford’s ongoing AI investigation on this blog, I decided try. I am glad I did. I came away from this experience very bullish on AI as an assistant for writing letters of recommendation.

In my inbox sat four requests for a variety of different letters of recommendation. One student, for example, was applying for a clerkship and another, a part-time student, to transfer to a school closer to home. I’ve written letters in these veins before, and my typical strategy is to pull up a similar letter for a similar student and use that as a template. I would then look up the details of each student’s performance, think about my interactions with the student, their relative performance in my classes, their unique strengths, and so forth. Then I’d set out to draft a letter. Perhaps it is unnecessary, but since I assume the same judges, hiring partners, and admissions committees will see multiple letters I write, I try not to copy and paste one student’s name and grades for another’s. It may be that I have almost the same things to say, but I hope a fresh letter better conveys my enthusiasm for the student I am recommending. And that’s where the tedium lies. It is also where, as it turns out, ChatGPT is a big help.

 

How are Law Students Using Claude and Other Large Language Models?

Take Reddit posts with a grain of salt, because their veracity can't be confirmed, but here's a recent post that may be of interest to those of us thinking about how to use AI in law school teaching/learning:

The screenshot may be hard to read, so here is the transcription:

I used Claude to get an A in a law school class that I never once paid attention to or studied for. I fed quality materials like old outlines and others' class notes into Claude, crafted the right prompts, and rocked that A without understanding a single thing taught in the class. Learn this approach, and you can cruise to top grades without ever cracking a textbook. But it definitely takes a certain kind of guts to pull this sh_t off.

To be sure, because of concerns about the use of AI to prepare exam answers, some professors who have given take-home exams in the past will consider switching (or have already switched) to in-class exams.

But is use of large language models by law students always a bad idea? One user responded to the post above with a suggestion:

Transcription:

A better approach is to personalize your learning about giving Claude the materials, and having it come up with new questions/answers based on the material. Try answering the questions yourself first. For stuff you got wrong, have it come up with related concepts. LLMs are amazing learning accelerators.

That seems like sound advice to me. I'm currently experimenting with having Claude generate new hypotheticals that I can use to supplement the 100+ practice questions I already give my students when teaching the basic Income Tax class.

I'd love to hear from other profs (and students) about how they are using Claude, Chat GPT, etc. to teach/learn course material.

Using Consensus AI App to Find Scholarly Legal Literature (No Hallucinations)

Chat GPT has made available via its free version (3.5) access to a custom GPT called Consensus that I am finding useful for locating scholarly articles that can give me a quick sense of scholarly approaches to pinpoint. Unfortunately, Consensus isn't linked yet to HeinOnLine/Westlaw/Lexis, so isn't quite at its full potential for legal scholarship, but it still can get a researcher pointed in the right direction, especially for interdisciplinary questions.

How to access Consensus? Within Chat GPT, choose "Explore GPTs" from the left-hand menu. Then, choose the "Research and Analysis" category under the search bar. Then navigate to the Consensus app, which is circled in red below (it looks like a "C" made out of two blue-ish colors):

What is Consensus? It's a search engine, like the ones most researchers are familiar with (the company's description is here; I have no relationship with Consensus except as a user). What makes it more helpful than, say, a search in Google Scholar is that Consensus combines the power of a large language model and the 200 million+ research papers in the Semantic Scholar database. Semantic Scholar includes materials from Cambridge University Press, U Chicago Press, and several databases that are mostly science-oriented. It also claims links with SSRN, but so far, I have found that Consensus is not super-reliable yet for scouring legal scholarship that I know for sure is posted on SSRN. Hopefully that will be fixed soon.

To test out Consensus, I asked a question to which I already knew the answer.  Because I've written about the topic, I could better evaluate the quality of the search results. Details after the fold.

Making Good Presentations at Academic Legal Conferences

As I continue to experiment with ways that artificial technology can be used to enhance legal research and writing, I've encountered several custom GPTs that purport to convert notes, paper drafts, etc. and convert them into PowerPoint presentations. Ones I've tried so far are by slidesgpt.com and studyx.ai. (I have no relationship with either, other than as an experimental user.)

I

So far, I have been impressed with the GPTs' ability to extract important information from my inputs, but the design of the slides is generic and repetitive. Perhaps with more incremental prompting, I'll get better results.

In the meantime, with a few conferences coming up soon, I've rediscovered this good advice from Philip E. Bourne (UVA, Data Science): Ten Simple Rules for Making Good Oral Presentations. Listed after the fold, the rules are pretty simple on paper, but not so easy for many of us law profs (myself included) to follow.

Using AI to Create Customized PowerPoint Images

The following is a guest post by Professor Rick Bales (Ohio Northern):

I use PowerPoint more than I used to. I think the visual stimulus helps keep students engaged, and the text prompts mean that I can ditch my notes without getting sidetracked. But I dislike text-heavy slides – I want the text to cover the highlights, and the rest of the slide to help engage the students. That requires some visual imagery.

Until recently, my images usually came from scouring the web. But for many abstract legal topics, nothing from the web quite fits, or sometimes it fits but the resolution is too low to look good on a projected slide. I’ve tried using PowerPoint’s drawing tools to create original material, but I’m not very good at it and it takes a lot of time to create just one good slide.

Over the last six months or so, I’ve been using DALL-E (the image creator that comes with ChatGPT4) to custom-design images for my slides. The main advantage to this approach is that I usually can customize my slide for its intended purpose. For example, here’s an image I used in my Employment Law course to illustrate a case about EMTs at Sacred Heart Hospital who thought they should be getting paid for their on-call time:

But not everything is customizable. I couldn’t get DALL-E to give me anything interesting to illustrate the “reasonable rate” for calculating overtime. And tort cases have been a challenge, since the DALL-E algorithm apparently has been instructed not to create images of people getting shot or maimed or killed.

Even when DALL-E can create the image I’m after, sometimes there are missteps. Words that appear in an image are likely to be misspelled. No amount of follow-up prompting was able to get DALL-E to understand that I wanted arrows pointed away from an object in an image rather than toward it. And sometimes, items in an image are just “off”, as in the image below, illustrating a workers’ compensation case in which an employee crashed his employer’s pickup truck into a tree while he was trying to retrieve a pack of cigarettes he had dropped on the floorboard. I too might have hit the tree too if it had been growing in the middle of the road!

On the upside, images like these keep students engaged. Students enjoy studying each of my slides to see if AI has gone rogue. I consider it a win any time my students are doing something in class other than watching cute puppy videos or shopping online.

My prompts tend to be short and descriptive. If I don’t get what I want after about 10 minutes of trying I might try scouring the web, and otherwise I’ll just go without an image on that slide. I don’t claim any expertise on creating these prompts. If you have suggestions on drafting prompts, or using other AI image-generators, or just generally on using AI to create better slides, please send them to me – I will compile and share in a follow-up post.

-Rick Bales

Will Claude AI Revolutionize the Law Review Article Selection Process?

I continue my experimentation with various large language models and AI tools to understand how they might be used to support legal scholarship. Lately, I've been playing around with Claude, a rival of Chat GPT.  (I have no relationship with either company.) It's free to use Claude's entry-level model. I have not paid for the upgraded version, which costs approximately $20/month.  ReadWrite has an informative review of Claude here.

I recently asked Claude to "summarize and evaluate the strengths and weaknesses" of the abstract and introduction of one of my accepted-but-not-yet-published articles, Taxing Sugar Babies, Minn. L. Rev. (forthcoming 2024). (For whatever it's worth, I'm trying to promote my paper per se, but because I wrote it, I can more readily determine how accurate Claude's results are.)

Results after the fold. TLDR; Claude's summary was pretty good and its evaluation of the weaknesses of the abstract and intro were fair. I suspect that student law review editors (either with or without being prompted to do so by Scholastica, etc.) will be integrating AI into the article submission process in the near future.

Revising Legal Writing with AI: What Hemingway Makes of the Pink Tax

One of my sabbatical projects this Spring 2024 semester has been training myself on a variety of new AI tools and exploring their applicability for legal research and writing, although not designed for those tasks. I signed up for a trial of the Hemingway Editor (free to use online + modest charge for download of desktop version; I have no relationship with the company).

I was interested to see what the app would make of legal writing, so I experimented with abstract of one of my recently published articles, Pink Tax and Other Tropes, 24 Yale J.L. & Feminism 87 (2024). I asked Hemingway to suggest a streamlined version. Results after the fold. TLDR; I'm not sure I love the edits.

Customizing Chat GPT to Write Like a Law Professor?

Like many others in law schools and beyond, I have been experimenting with Chat GPT to better understand its capabilities and limitations. From an online seminar led by Mushtaq Bilal (Literature, University of Southern Denmark), I learned that about customizing ChatGPT (clicking on your name in the lower left corner of the home screen-->click "Customize ChatGPT"). You can provide information about your areas of academic interest and how you want ChatGPT to respond (i.e., use a formal, academic tone; here's what I expect you to address if I ask for a critique of a paragraph). My anecdotal experience so far is that providing these customized instruction makes ChatGPT a more accurate source of assistance.

Separate and apart from availing oneself of the "Customize ChatGPT" screen, one can also ask ChatGPT to remember certain things, including what you have written in the past. That way, ChatGPT has a better sense of the tone, output, etc. you may be expecting (more explanation here).

I recently experimented by uploading to ChatGPT four of my published law review articles and asking it to remember that I am the author of those works.  I then asked ChatGPT to "write a paragraph about balloons in the style of Bridget J. Crawford." Here were the results:

 

I have never written about balloons before, and I haven't written anything substantial about environmental issues (or anything about consumer safety). So it is difficult to say if this is written in my "style." But I do think that ChatGPT's output in response to this prompt has some noteworthy qualities that seem "law prof-ish" to me: a coherent thesis statement, a mention of variations in legal framework, the reference to regulations, and linking to other issues ("a microcosm for examining law legal systems can evolve").  I'm interested in continuing to experiment--cautiously, ethically--with this technology.

Has anyone else experimented with custom instructions? Any tips for what is working well (or poorly) at this stage?

Drexel Kline Law Seeks Law & Technology Postdoctoral Fellow

The Drexel University Kline School of Law’s Center for Law and Transformational Technology invites applications for a two-year postdoctoral fellow, beginning September 1, 2022, with an emphasis on sociotechnical research in the areas of law, regulation, and governance at the intersection of law and technology, and the development of new technologies. Applicants from all disciplines and methodological approaches are welcome to apply.

The Center for Law and Transformational Technology was created in 2021 with a seed grant from the Green Family Foundation to bring together key stakeholders, including legal and technological scholars, industry leaders, activists, regulators, and other experts from across the country and the world. The Center supports concrete efforts to gauge the legal impact of transformational technology, promote critical conversations about the role of law in technological change and bring legal experts into interdisciplinary conversations with researchers and developers at all stages of design and development. Working across industries, the Center helps educate lawyers to play a positive role navigating a technological landscape we cannot yet imagine. By facilitating international and multi-disciplinary partnerships, the Center seeks to contribute to a future in which technology moves forward hand-in-hand with society.

The Fellow will also have the opportunity to collaborate with faculty and participate in workshops and symposia from across the University. Situated in the heart of Philadelphia, Kline Law and the Center for Law and Transformational Technology offer exciting opportunities to engage with the region’s intellectual and innovation communities.

The Fellow will work with law faculty and faculty in cognate disciplines who work with the Center. The Fellow is expected to conduct original research, teach at least one class each year (for JD students or undergraduates studying in the Law School’s newly established B.A. in Law program), and support Center activities including symposia, outreach to scholars in varied disciplines, and a student research fellow program. The Fellow is expected to participate fully in the intellectual life of the Center and of Kline Law.

The successful candidate will have obtained a Ph.D in a related field, a JD or an SJD.  Applicants should have experience and an active scholarly agenda centering on the intersection of regulation, law and/or governance with respect to emerging technologies. Preference will be given to applicants who have published in this field. 

Details and application instructions are here. 

The Law School Industry - Law and Technology @ US Law Schools

Digital technologies are an increasingly important phenomena in higher education and professional industries. Law schools have started to participate in rethinking how to train the next generation workforce/citizens at the interface of governance/technology. This post includes links to quantitative data for folks interested to learn more about the current law/tech situation in US legal higher education, and provides a few initial thoughts about the data. 

Here is an excel sheet formatted study with a wide range of data on current US law school efforts around 'Law and Technology': Download US Law and Technology Higher Education Law Schools - Haskell:Fish In addition, for an alternative visualisation for some of the information, see: https://app.flourish.studio/visualisation/2736840 

Note: The study was developed along with a PhD student, Jessica L. Fish, and funded by the University of Manchester Law School and its Law and Technology Initiative. Scholarship based partly on this data will be rolled out in the coming year, but the information provided is meant to be open sources/common for any colleagues however useful. 

A Few Thoughts... (see below)

More Online Teaching Resources

To add to the current compilation of materials on how to move to remote teaching, I wanted to share the following SSRN links.

The first was posted this week by Prof Agnieszka McPeak (Duquesne) and is a straightforward, simple best practices guide to moving things online. Six pages, easy to read and great advice:

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3553094

The second is a piece I wrote last year which is longer and focuses more on the costs and benefits of law schools moving to online teaching, but it also includes some practical observations on larger and smaller group teaching online:

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3491427

Hopefully some of those struggling to get online quickly may make use of these pieces. I highly recommend Prof McPeak's thoughts.

 

The Great Mass Online Teaching Experiment

For many years, I've been fascinated by the possibilities of online teaching. Things like MOOCs and flipped classrooms seem to offer real opportunities to teach in new and better ways. All the same, I've been busy with other things, and actually making the leap to offering online instruction never quite made the list.

Then came Covid-19. My students are sequestered in apartments all across China, and the Peking University School of Transnational Law campus in Shenzhen is empty. I've chosen not to return to China after the spring festival, but have been hunkered down in Chiang Mai, Thailand, waiting for the situation to gain some clarity. The clarity that has gradually emerged is that we aren't going to be teaching face-to-face for at least a while, and we're taking all our classes online.

In going online, we're joining all kinds of schools all over China. I've got a cousin who has retired from K12 teaching in Canada and is running the international section of a large school on the east coast. They've taken all high school classes online, along with most similar schools. Teachers report that students locked down in house arrest tend to be highly prepared. At an expat event in Chiang Mai, I met two kindergarten teachers from an international school in another part of China. Although the ministry of education doesn't support online teaching for kids that young, in response to parental demands they have been teaching children as young as three years old online. While hugs are hard to deliver, they can record sing-a-long songs or record readings of English books and lessons, and comment on the responses the children's parents sent back to them. Possibly hundreds of millions of students have been shifted to online instruction in China, something that has not happened at that scale in the traditional educational environment anywhere before.

My students, fortunately, are a lot more self-sufficient and able than three-year-olds. It's been a challenge to move everything online, but it seems to be working. Here are some quick thoughts. As my experience deepens, I might be back with more thoughts.

Platforms. There seem to be a lot of good platforms out there. We are using Zoom, in both its webinar (better for classes over 100 in size) and meeting (better for smaller classes) sizes. I find Zoom pretty easy to use, although with each class I find some features and shortcuts I hadn't noticed before that make it still easier. I am able to call on students by name, put them on camera, share their video with everyone, and in general carry on pretty much as we do in a regular classroom. Our budgets tend to be pretty limited, and we aren't using some of the complementary software that enables things like seamlessly moving to pre-recorded snippets. So far, that seems to be ok.

Bandwidth. As you would expect, bandwidth is an issue, both for the students and the teachers. This week in order to renew my Thai visa I took a side trip to Luang Prabang in Laos, where the internet just isn't robust enough for online teaching. To deal with that, I went asynchronous for one class, recording a lecture before I came over. Students are also differently positioned, and some students experience issues with audio quality or dropped connections. To ameliorate, we are recording all classes, and students can download and listen to any difficult portions at their leisure. I find 20 megabits per second connections more than adequate and five megabits per second too low for full functionality. (Type internet speed test into Google to get a test that will tell you your connection speed).

Disadvantages. Some of the issues involved in moving online are obvious, but others less so. For example, in our situation, library reserve - which I use heavily - is not available. In some courses, we have copies of the basic text or core readings for everyone available in the library. That doesn't work when students are scattered. For example, in my Law and Innovation class we have enough copies of the Susskind's The Future of the Professions to go around when we are all in Shenzhen, as well as enough copies of Thomas Morgan's PR book to go around when we cover specific ethical issues in the legal profession. We have no way of getting to those to students all over China. That requires adjustments in the course materials as a small subset of my students can buy whatever book I mention and others are financially constrained, so simply telling them to go to Amazon or JD is not a good solution. In some cases, the only solution has been for me to write substitute materials, which has been a huge time commitment.

You can't see the classroom as you talk in the same way you see them in a live classroom, so the skills you've developed in reading those cues aren't on point. 

There are lead time and set up issues. I use PowerPoint slides, and I've found it helpful to the students to circulate those before class so they can refer to them in a separate window. That requires having them locked down a day or so ahead of time. Online meeting links also need to be generated and circulated. I've also been designating a list of on-call students, something I don't always do in person, so that the students can be sure their cameras and microphones are ready to go. 

One thing I've not sorted out is assessment in my Civ Pro class. I typically give an exam with multiple choice and essay components. I'm not sure how I am going to do that if we are not back on campus by exam time. There are software solutions that will lock student's computers down, but we don't have them. For self-assessment, I'm using the same online quizzes on TWEN that I've used for years.

Advantages. Some of the advantages have been surprising as well. Zoom allows students to ask questions anonymously in the Q&A. This allows students to put things out there without losing face or seeming stupid, and it actually generates way more student commentary and questions than I get in a live class. I get so many comments in my Civ Pro class that keeping the flow of class moving ahead has become a bit of an issue.

Another advantage that I hadn't thought about in advance is that making the classroom non-physical opens up the world of guest panelists and speakers. I'm busy trying to pull in thought leaders and people with on-point experience to take a few minutes in both my Civ Pro and Law and Innovation classes. For example, in Law and Innovation, which has a large PR component, I'm bringing in a lawyer from Kentucky to talk about the devastation that was visited on the local community around Butcher's Hollow (yes, that Butcher's Hollow) when a corrupt local lawyer entered into a bribery deal with a social security judge to have his clients all approved as disabled. The upshot was that the Social Security Administration cut off benefits for all the clients of the corrupt lawyer, most of whom were in fact disabled, and who were neither aware of nor involved in the fraud, leaving many of them in perilous straits. I'm hoping that having live in the classroom someone who was involved in cleaning up a tragic situation caused by lawyer dishonesty will make the real-life consequences of lawyer misconduct more palpable. (By the way - feel free to volunteer to be a guest speaker).

My students, I think, are enjoying having all classes recorded and available for download. Setting aside technical problems, if they drift or lose focus for a moment, they can replay the section that did not quite make sense. If it does not make sense because I did not make sense, they can bring it up in online office hours. (Having every word recorded and distributed might be an issue for those of you with aspirations to sit on the Supreme Court or even just good reputations to protect, and if that's the case you can choose not to enable the recording feature.)

One obvious benefit of teaching online is that you can become a digital nomad and be someplace new and different. The attached photo is from my temporary office at a riverside cafe a few days ago.

As this goes forward, and I am only four classes in, I may post more. Comments on online teaching techniques are welcome. Comments reflecting your personal conspiracy theories on the origins and dispersion of the virus that causes Covid-19 will be deleted.