Honored to have been selected as a Schmidt Futures AI2050 Senior Fellow -- all working on the hard problems of AI. I'll be devoting my fellowship to the challenge of how to build normative infrastructure for AI alignment. gillian.hadfield
@threads
.net
Very excited about this new opportunity as inaugural Director of the Schwartz Reisman Institute for Technology and Society
@UofT
. Looking forward to working with folks in AI and across social sciences and humanities.
I’m super excited to say I’ll be joining the faculties of the brand new DC-based School of Government and Policy and School of Engineering (CS) at Johns Hopkins starting July 1, 2024. An amazing opportunity to amplify my work on AI policy and alignment /1
Happy to share that I won the Mundell Medal for Legal Writing (writing on legal regulation & reform including Rules for a Flat World) along with David Tanovich (writing on race & criminal justice including The Colour of Justice)
@UTLaw
@UofT
@WindsorLaw
Very sad to hear about the sudden loss of Peter Eckersley--fitting that this was his last tweet. A beautiful mind, committed to deep original thought about AI. A rare person in AI safety who engaged deeply on economic theory and even parenting principles
The legal world lost a giant yesterday: Deborah Rhode, the woman who did more than any of us to address the A2J crisis—her work has been central to this effort for decades. A treasured coauthor, activist, and friend.
I don't know if we can or will build massively capable autonomous AI with long-term planning abilities. But I do know this: whether such AI gets built and deployed should be decided by us through governments, not markets. We make the case
Calling all social scientists, humanists, and AI researchers working on the social and ethical dimensions of AI: 3rd AI Ethics and Society Conference will be co-located with AAAI in New York Feb 7-8 2020. Call for papers is up; submission deadline Nov 4.
Glad to finally have this paper with
@jackclarkSF
up--presenting our take on the landscape of AI governance, the critical democratic and technical deficits in current efforts, and our proposal for regulatory technologies and regulatory markets to respond
It’s really time to admit that we have allowed tremendously complex legal processes to develop that exploit the fact that the vast majority of people cannot manage tremendously complex legal processes
Save the dates! AI Ethics and Society 2021 Conference will be held virtually May 19-20-21. Submission deadline January 15, 2021. We invite submissions across AI, social science and humanities. Join the work to build safe and beneficial AI.
Utah S. Ct. is about to be the leader in North America for
#A2J
--they've endorsed the most progressive plan I've seen for regulation of legal services--on track to eliminate access-killing rules against fee-sharing, entity practice, and innovative business models. See here:
Great Vancouver Sun piece today really captures the outrage: lawyers can fix the
#A2J
crisis but still refuse to do so. Let’s get the public thumping tables for change. People deserve affordable legal help.
@UTLaw
This may seem arcane but it's a big deal: European court holds that standards created by private standard-setting bodies, eg. to implement EU AI Act, must be available for free (not sold as is status quo). Rule of law, transparency and all that good stuff.
SO excited to have played a part in this historic move. Utah has acted where so many other states have merely formed yet another study group to solve the unconscionable lack of access to reasonably priced legal help that lawyer (and court!) regulation has created
You know you've arrived when the incomparable
@margarethagan
illustrates your pitch to CA State Bar about why they need to change the regulatory model of law to solve the
#A2J
crisis!
Heads up folks:Utah Supreme Court just did what no other court or regulator has ever done: APPROVED a major shift to an evidence-based approach to legal regulation that can transform the landscape of legal services for all. Not "let's study more" just "let's do this." Bravo!
Utah proves itself a major leader in the regulatory reform we need for
#A2J
and
#legalinnovation
. A superb report (if I do say so myself as a contributor..)
I'm happy to announce I've started to work with
@OpenAI
as a Senior Policy Advisor -- seeking to support their mission to ensure that the path to artificial general intelligence benefits all of humanity. Coming soon-AI policy ideas we can implement now
Another big step for
#A2J
. Arizona Task Force recommends eliminating Rule 5.4 which prohibits lawyers from working for or with other professionals and businesses and urges immediate evaluation of building regulator for entities other than lawyers, following Utah lead. Bravo!
My TED talk asks: How do we build trust in the age of Big Tech? Answer: current regulatory approaches won't work; we need to get as serious about innovating reg tech & rules as we are about technology
Arizona Task Force votes unanimously to recommend abolishing Rule 5.4 which prohibits lawyers from using business models essential to reducing costs and improving quality (corporate form and profit/revenue-sharing contracts). About time. .
#A2J
.
Our latest
@ScienceMagazine
urging greater R&D efforts in AI governance and safety--even if you think super-powerful systems are far off, think about how much research we need and how long it takes to build new societal consensus and legal infrastructure
The demand for legal work is exploding...not shrinking. The demand for the existing type of legal work that’s being supplied in the market, that’s what’s shrinking.
The amazing state of Utah has announced its proposal to make the most thoroughgoing changes to the regulation of legal markets, ever. Sandbox already open for covid related initiatives. Comment and let them know this is awesome! Finally true
#A2J
progress!
Demand for legal tools to manage social and economic relationship is not dropping, it is ballooning. What’s dropping is demand for what our current legal infrastructure is supplying.
#legalinnovation
"Almost half of U.S. law firms surveyed said that they could not reach their billable hour targets last year and around the same amount admitted that their partners are 'underutilized'"
- What are folks opinions on this?
#lawfirms
#law
@akbarpour_
Ah. Paul said something similar to me when I was his student. I had come to discuss my work and said something like “I think I can show X”. And he said, “we don’t build models to show things, we build models to discover things.” I have been guided by this throughout my career.
What terrific news out of Arizona--their Supreme Court has approved rules changes allowing lawyers and legal businesses to operate more efficiently and effectively--joining the leadership of
@UtahReformer
and his colleagues in Utah! Excellent news!
Wait it gets better: Utah just announced that new legal providers offering low or no fee services for people facing covid related legal problems in housing, employment, small business etc can get early approval to enter Utah sandbox and deliver services
I'm looking forward to giving a
@TEDx
@TEDxToronto
talk (virtual! from my basement!) next Thursday March 4 (7 pm Eastern) on the challenge of reining in big tech and why we need to be as innovative about our rules as our technologies. Tickets here:
Utah proves itself a major leader in the regulatory reform we need for
#A2J
and
#legalinnovation
. A superb report (if I do say so myself as a contributor..)
Read this paper! Studying human collective behavior should be a 'crisis discipline' given speed of how technology transforming human interaction. Aligns well with my thinking on human normative systems (the way we make rules) as complex adaptive systems
Go Becky! A clear call for critical change from a MacArthur Genius who has done the hard work of proving that most people can’t get legal help and the lawyers’ monopoly is why
This is not supposed to happen: the estimates of those without meaningful access to law has gone UP globally from 4 to 5 billion. Opinion: Most people have no access to basic justice. But the fix is easier than you think
My advice to other women who want to get involved in legal tech?
Don’t start with the tech. Start with the underlying challenges people and businesses are facing in their lives
@ltrc
#WOLTC2018
We urgently need innovative approaches to law that grapple with the fundamental transformation that we are experiencing in the nature of how business gets done and how economies operate
#GlobalLegalHack
#GLH2018
#HadfieldChallenges
1/2 Utah, the first in the nation to reform their regulation of lawyers to promote access to justice, has now also become first to create an Office of Fairness and Accountability to study the data to identify
#systemicracism
and other bias in courts.
Important new paper exploring whether a technique of human feedback can help to align powerful language models with a set of human preferences/intent--great opportunity for complementary work from social scientists thinking about collective choice and political economy
We've trained GPT-3 to be more aligned with what humans want: The new InstructGPT models are better at following human intent than a 100x larger model, while also improving safety and truthfulness.
Want to help lead our Institute to do research and build human-centred solutions to make sure advanced technologies like AI are safe, fair and inclusive? We're hiring: Director, Strategy & Operations, Schwartz Reisman Institute for Technology and Society
AI that is understandable is important. AI that is justifiable is essential. They’re not necessarily the same thing and we need more work on the latter, not just the former
Very pleased to say that the audiobook version of Rules for a Flat World is now available on Audible! Prologue read by yours truly, and the rest by RH Thomson, award-winning Canadian actor who does a terrific job and is a pleasure to listen to.
Folks who understand how important this is and what an achievement—please help balance out the negativity from lawyers who don’t see that and comment on the proposal. Your comment can be short and sweet—this is a much needed and long overdue change that will benefit many
#A2J
And here we go: Utah proposes non-lawyer law firm ownership, end to restrictions on fee-splitting with non-lawyers, and creation of a new Office of Legal Services Innovation to oversee a regulatory sandbox. Comments open through July 23.
Just learned about a new LLM-powered legal advice system aimed at consumers-not lawyers or large corporates. is based in Toronto but launched only in the UK (because legal tech for consumers is not illegal there!) Great update here
Super pleased to announce that the
@StateBarCA
Board of Trustees voted by a large majority to move forward with pursuing a regulatory sandbox that will eventually be able to increase access to legal services in California.
Glad to have been able to contribute to this project
@OpenAI
—recommending steps for industry for building cooperation in framework for responsible AI development and framing the case for regulatory solutions as well
Really excited that this important paper/project, led by
@AmandaAskell
with contributions from myself and
@ghadfield
, is finally out! Looking forward to people's thoughts on our paper and the issues more generally.
Here, from me, Tino Cuéllar and
@timoreilly
: a concrete, measured proposal for what we should do, now, to build the regulatory environment for generative AI: start with a national registry
@jsench
Hah! I’ve been putting a sentence “please send me an email if you read this far” into my syllabi for a few years—I get few emails! Your approach sounds much more effective!
Good idea: auditing algorithms as a service. Better idea—spin out to independent company. Even better—regulate this service so it meets democratically set goals as
@jackclarkSF
and I propose. via
@wired
Big news! We're excited to announce our partnership with
@_AIGlobal
in collaboration with the
@wef
to build a universally recognized framework that validates
#AI
tools and technologies as responsible, trustworthy, ethical, and fair.
Read more:
#TorontoSRI
My standard pitch to law students when I’m trying to convince them that my focus on problem solving and judgment and teams really IS what they’re supposed to be learning
I'll be in London June 18 to deliver the UCL Centre for Ethics and Law Biennial Lecture--speaking on the challenge of regulating artificial intelligence. Join me if you are in that great city! Reserve a seat here
Great to see this report showing gains from legal regulatory reform model we built in Utah with
@LucyEdmondB
. And sad to see lawyers still saying, shameless, that reform is bad because it is "shifting profits of the legal industry to nonlawyers." Law doesn't belong to lawyers.
The Utah model of reform allowing nonlawyers to offer legal services could be “critical” to serving people who can’t afford them, according to a Stanford Law School study published Tuesday.
#accesstojustice
@mreynoldswriter
Learn more about how ChatGPT is being built and in particular the role of human feedback and how to participate in shaping the future of those processes and these technologies.
COVID-19 is exposing everything that is broken and the abysmal state of access to justice around the world is right up there. It’s a time when people need help in our law-thick world—to manage employment, family, small business, health care etc. legal jungles. Change needed NOW
With our partners in the
@SDG16Plus
alliance, we joined professionals from around the world come together for a first briefing on what the
#COVID
ー19 pandemic means for people's
#accesstojustice
. Read a vital new briefing on
#JusticeForAll
in a pandemic:
@DBRodriguez5
@ABAInnovation
@wihender
@CrispinPassmore
@CalBarTrustee
Can we all stop quoting that ridiculous headline?! 250 lawyers complained. Of 170,000 lawyers in the state. And 17 million citizens with legal problems, the vast majority of whom get no legal help thanks to the rules the CA task force is trying to change. That’s not “swamped”.
I'm stoked--if we do this right, I think we invent whole new fields of research, making AI research smarter about how humans work and social sciences smarter about AI, and as a result making sure the future is better for all.
"It’s exciting to imagine how the new Schwartz Reisman Institute for Technology & Society could encourage scholars to think differently all across the university about the challenges and opportunities posed by new technologies," says
@UTLaw
's
@ghadfield
@AndrewArruda
@StateBarCA
What’s especially shameful is that lawyers do not even try to hide that they are using their political clout to protect themselves at the expense of the 80-90% who have no access to legal help.They will sneer and say where’s your evidence but threaten those who would gather data
Not often an academic publication sells out within 36 hours but editors Linc Caplan, Lance Liebman and Becky Sandefur have done it with this issue of
#Daedalus
on
#accesstojustice
My contribution on "more markets, more justice" Access is free (of course!)
A great idea from
@djjr
for everyone working from home or teaching online classes: invite your students or co-workers to "tea" daily at a set time over videochat--virtual break-room. Isolation is not good for anyone.
"the civil justice system in America is failing. You can measure this failure in many ways...86% of civil legal problems either get inadequate legal help or no legal help at all. We need solutions that are commensurate with the crisis at hand." -
@LSCtweets
James Sandman
#A2J
Woohoo!
@dhadfieldmenell
and I have had TWO of our papers accepted for presentation at the
@RealAAAI
affiliated AI Ethics and Society Conference in Honolulu Jan 27-28! Incomplete contracts and AI Alignment and Silly Rules .
Advice for zoom/google/skype/teams meetings: turn on your camera! My team
@TorontoSRI
@UofT
has been having twice daily video checkins and they are a godsend. Don’t spend the day talking to black screens! See faces! Know that you must get up and shower and dress! Stay connected!
Thoughtful reaction to my paper with
@OpenAI
@jackclarkSF
paper—Still very much a work in progress but we’ve been delighted by extensive interest in figuring out this model. Presenting at
@Nber
AI Econ in Sept. Goal design is a good point. As is reinventing government oversight
Excited to share a new post on evaluating a regulatory framework for oversight of AI.
@jackclarkSF
,
@ghadfield
, your feedback would be much appreciated.
I *really* enjoyed doing this long-form podcast with two very smart people who knew their stuff and gave me a chance to talk about everything from cultural evolution to multiagent reinforcement learning, and the regulatory innovation we need to address AI risks and opportunities
→ Why did humans invent law?
→ How can laws keep up with the pace of technological change, and how might "regulatory markets" help?
→ And what is the value of "silly" rules for our normative infrastructure?
Listen to
@ghadfield
–
Pretty awesome—trading an algorithm to recognize legal issues in the way ordinary people talk about their problems. Stanford and Suffolk Create Game to Help Drive Access to Justice | LawSites
A nice overview of a longer paper for those looking to know what problems researchers interested in AI alignment (getting AI to do what you want it to do) A Summary of Concrete Problems in AI Safety
My hero: Jim Sandman, radicalized by nine years leading the Legal Services Corporation about the urgent need to reform the rules of legal practice to address the crushing access to justice gap: How to navigate parochial and self—interested concerns (189)
The first step to building better law is to peel away a fundamental fallacy: that the building of legal systems is a task that should be left to legal experts
California should NOT go backwards on opening up legal markets for access to justice. There is just no way retrenching is explained by anything other than lawyer self-interest exerted as budget threat on the Court
1/7: Dear
@SenatorUmberg
&
@AsmMarkStone
, I read your December 7th letter and signed a reply. I believe you are acting in good faith. The California Justice Gap Study suggests at least a third of your constituents in your respective areas faced legal problems without any help.
Hello world!🌎We're here to foster collaborative research across the sciences, social sciences, and humanities to build real and practical solutions that ensure technology works for a more just, equitable, and productive society. Follow us on our journey!
#TorontoSRI
#LetsSolveIt
Good catch by
@jackclarkSF
.The weakness of research design in AI is a real problem--too many folks *believe* AI results uncritically.Machine trained to detect deceptive gaits of students ASKED to walk around "deceptively". Oy. Would never make it past social science referees
Full show available at this link. A great discussion ranging from the future of AI to misinformation, plagiarism, and the potential for ChatGPT to provide legal advice! Not to mention how children learn and the history of technologies and how law and regulation adapts…
Very touched by what
@wihender
has to say about what we accomplished in Utah on behalf of change and
#A2J
—and what we still have to do. Dropping the Rock: three examples (112) | Legal Evolution
It’s basic economics that lawyers should take to heart—when you make your service more affordable, more people can buy it and the total value to be shared goes up. More markets, more users, more work. Demand curves slope down!
Join Us in Unlocking Legal Regulation: More Legal Service Providers, More Access to Justice, and a More Sustainable Legal Profession:
#reimaginelaw
#unlocklegalregulation
And the paperback, with a foreword addressing AI, is out (sort of—locked into a warehouse somewhere waiting for the world to reopen—pre-order on Amazon available)
KCL CSGS: Centre For Study of Governance & Society
Law Society of Ontario approves regulatory sandbox for legal tech (that now makes two of Canada's three largest provinces, if you're counting)
@journo_dale
@jordan_law21
@ghadfield
Nice to hear! Thanks! Paperback version available as well with foreword on the challenge of regulating AI given all that stuff about institutions, cultural evolution, cooperation and history...
@DavidSkarbek
By
@ghadfield
- Reading it now (on a recommendation). nominally about law but actually about about institutions, cooperation, cultural evolution etc. It’s really interesting & lots of lovely historical examples.
1) Do I misread program
#asa19
? Maybe 6 papers on AI, mostly ”algorithms discriminate” & “robot take jobs,” a few more on ML, mostly methodology. Nothing on normativity, alignment, society +AGI, law + AI, complex orgs as model for AI. So much rich untilled soil.
If you want *the* best take on AI policy, and the urgency of action, and why we are failing spectacularly to take seriously the massive challenge to our socio-economic systems, read the whole thing.
"Ambitious as they are, no legal entrepreneur has similarly revamped the economics of legal services." Why we need to rethink the rules like SpaceX
#legalinnovation