12/4/2024
AI-Related Wizards
A View into the Generative AI Legal Landscape 2024
1st-Build Track ➲ LLM Submissions
Continuation Track ➲ LLM Submissions
Technology & Applications of Computational Contracts in Insurance & Finance Workshop
Practical Applications ➲ The CodeX Insurance Analyst (CIA)
The 2024 CodeX FutureLaw Conference
SALI Alliance ➲ Modern Legal Industry Standards
Using AI for Simulated Legal Training
David Engstrom ➲ Closing Keynote
Norm.ai ➲ Regulatory AI Agents
CodeX [Stanford Center for Legal Informatics] describe their 11th annual FutureLaw Conference as follows:
"A deep dive into the transformative impact of technology on the legal profession, the justice system, and societal regulations.
This year, in light of exponential advances in LegalTech and computational law, we go beyond the surface to reflect on the questions that we hesitate to voice. That is, are we ready to redefine the role of law and its existing architecture?
CodeX FutureLaw 2024 is the nexus for the forward-thinkers, the risk-takers, the innovators. Academics, entrepreneurs, lawyers, investors, policymakers, and engineers will converge to enlighten, debate, and together shape the discourse on the future of legal."
The preliminary events in the lead up to the CodeX FutureLaw2024 Conference include the following:
⚖️ A full day #nocode introduction to generative AI for all members of the Stanford community;
⚖️ A lively Hackathon community Discord chat channel (Circa 500+ invited members at the time I joined a few days before the Hackathon) where teams were encouraged to introduce themselves and their projects, organize meetups, form teams, share resources, brainstorm ideas, volunteer for showcase lighting talks, etc.;
⚖️ The Hackathon; and
⚖️ The Computational Contracts - Insurance & Finance Workshop.
On or about 7 March 2024, a Stanford published academic paper by Dr. Megan Ma, Aparna Sinha, Ankit Tandon & Jennifer Richards provided "A View into the Generative AI Legal Landscape 2024".
I have taken the overview from the related blog article and asked AI to rewrite it in a professional and concise manner.
The rewritten overview generated by AI is extracted below:
"The LegalTech field has been revitalized by generative AI advancements, leading to a surge in AI-first startups and significant venture funding (~$700 million since early 2023).
Challenges include structural barriers and incumbents' data advantage.
Key points:
⚖️ Technical advancements are improving generative AI accuracy in legal applications.
⚖️ Structural hurdles in the legal industry moderate success of AI startups.
⚖️ Client-facing LegalTech market segmented into Research, Document Review, Litigation, with a total US market estimate of ~$13B in 2023.
⚖️ Incumbents play a key role in adopting generative AI.
⚖️ Future trends may include specialization in patent, IP, immigration, insurance, and regulatory compliance.
⚖️ Founders should consider industry dynamics and hurdles when introducing AI tech, focusing on less data-dependent areas or partnering with incumbents.
⚖️ Despite challenges, generative AI in LegalTech offers deep insights into legal practice.
For the full paper, click here."
Extracted from the Hackathon invite landing page:
"As artificial intelligence (AI) continues to revolutionize industries across the globe, the legal sector is no exception. LLMs, a foundation model capable of understanding and generating natural language text, have the potential to transform legal research, contract analysis, dispute resolution, and much more.
This hackathon hopes to build solutions that consider the safe and trustworthy integration of LLMs for legal work.
Join us for the third edition of the LLM x Law Hackathon!
We encourage you to think boldly, creatively, and across multiple dimensions!"
For the first time the Hackathon offered 2 tracks:
First-Build Track:
This is your typical hackathon style submission – show off your dev skills using the most cutting edge AI tools as you build & ship the best legal hack possible during the hackathon!
Continuation Track:
Bring your ongoing Legal Tech project and continue shaping it during the Hackathon. This track is ideal for showcasing your work and products you have built.
Whilst I was unable to attend the hackathon due to other commitments, as a participant in the Hackathon community discord channel I was able to view the hackathon team submissions.
It is evident from the diversity of ideas for AI based LLM powered solutions that there is no shortage of potential applications.
I have compiled the following non-exhaustive list of examples (mainly for my own benefit to gain insight into what I missed by not being there) but also to showcase the variety of both 1st-build track and continuation track hackathon team submissions.
For privacy reasons, other than providing the key contact person's LinkedIn profile link for the publicly announced winning submissions, by default, I have deleted team member names.
Please contact me If there are any updates to submissions I have missed or details or links to add for working prototypes or demo videos, or if team members would like to add their name & contact details.
The 4th LLM Hackathon "Hack_the_Law" will be held in Cambridge on June 23, 2024!
"Focused on empowering Veteran’s during the Disability Claims process, our team has designed a Disability Claims Assistant (DCA) and Web based experience.
Vet’s Claim is designed to empathetically explore the Veteran’s conditions, military history and their current quality of life.
Our product informs the Veteran about which of the 910 VA conditions they are applicable to apply for, the documentation and language that successful applications include.
Our tool also provides written documentation for their submission, their Doctor’s substantiation and any applicable witness documents.
The final output is a single PFD file for the VA that meets all application requirements."
"DueDilgent does a deep dive into key contracts to speed up M&A fact finding. It's fast, accurate and efficient. It's built to fact check business valuation assumptions against contracts provided by the seller."
"We train junior law firm associates with critical skills such as negotiation, communication, and redlining, ensuring their value and resilience in the evolving legal landscape.
Our product offers a cutting-edge training platform that provides immersive simulated negotiation scenarios.
The simulated negotiation will be through a Word document and in redline, and the AI will act as a law firm partner who is reviewing a junior's work."
"An AI that suggests and drafts objections to interrogatories during the discovery phase of civil suits."
"We leverage AI to quickly find prior art patents and publication to invalidate a patent."
Link: https://youtu.be/qRP2qfi7sHI?feature=shared
"Inspired by real-life struggles, VisaVantage was born out of the need to assist international students in navigating the U.S. legal system. A friend's ordeal with visa complexities and legal hurdles underscores the necessity for a dedicated platform."
"Facing 60,000 calls and 150,000 digital inquiries annually, the legal landscape for international students is fraught with challenges. VisaVantage, leveraging AI, simplifies this journey, offering clear, personalized legal guidance."
"VisaVantage reduces legal complexities, aiding compliance and empowering students with knowledge. Yet, the evolving nature of immigration laws poses a constant challenge, necessitating continuous AI adaptation."
"We invite partners to enhance VisaVantage’s capabilities, ensuring every international student can confidently navigate the U.S. legal system. Together, we can build a future where legal barriers are minimized, and educational aspirations are realized."
"An application that analyzes a home purchase agreement, and highlights high risk terms that deviate from standard home purchase contracts."
"Our solution creates Due Diligence reports of merger and acquisition agreements to streamline the work of in-house counsel."
"Judge the Judge uses LLMs to provide insights into judicial opinions and tendencies.
The tool is designed for legal professionals, researchers, and policymakers aiming to understand and predict judicial behavior regarding specific topics or case types."
Link: https://chat.openai.com/g/g-EJunP4HyT-judge-the-judge
"We're building an AI Assistant that provides 24/7 instant and personalized response to your questions regarding immigration process (specifically focus on OPT/STEM OPT/H1B procedure. Conversation memories are maintained."
"We create an AI-powered Patent Due Diligence Tool which can performa a claim-by-claim analysis of the patent.
When you input your invention description, the tool can create a spreadsheet containing feature-by-feature comparison of your idea and the most relevant existing patents."
"Tool that periodically visits all county websites and adds new laws to a database. Users can then perform semantic search to find out about key law implications relevant for them."
"We provide intelligent data compliance solutions for California’s emerging companies in the healthcare industry to streamline compliance procedures and reduce legal costs by automating preliminary compliance checking and flagging potential issues."
"Our project leverages AI to automate breach materiality assessment for SEC Form 8K reporting, bridging the gap between cybersecurity and legal teams, ensuring swift and compliant documentation of cybersecurity incidents, and reducing compliance risks for organizations."
"Junior Paralegal" is designed to greatly enhance a paralegal's efficiency when searching for relevant cases. This tool maintains all data internally, offering a semantic search that enables paralegals to complete their work ten times faster! It's an affordable solution that significantly speeds up your search."
"Agreewise is your legal contract assistant for attorneys to review agreements more efficiency. This reduces switching costs and improving operational efficiency."
"Aitana is a tool that quickly translates court speech into many languages, helping people who speak different languages understand everything in real-time. It's easy for anyone to use and helps make sure everyone is treated fairly in court by removing language barriers. This tool makes court better and more accessible for everyone involved."
"Non-lawyers and lawyers alike struggle to understand long, complex contracts. Agreedients boils contracts down to their essentials, enabling anyone to understand key dates, obligations, and financials at a glance. Results are much better than uploading a PDF contract to ChatGPT because legal domain knowledge is built into the tool at every step.
We use an Option Agreement to show how Agreedients can demystify stock plans for employees, both during the stressful negotiation steps and post-execution."
"Our tool parses legal briefs, detects citations, and automatically updates them to Bluebook-compliant formatting using contextual reading and LLM. We provide a co-pilot experience with auto-completion and smart suggestions, simplifying citation tasks for judges, lawyers, paralegals, and legal scholars."
"An AI tool that generates accurate summaries of events with timeline generation categorized by key topic and person. Simply upload documents to the tool that extracts key value as well as key figures and chat with Paralegal AI for streamlining paralegal's manual data preparation and organizational workload."
"We are building an AI assistant to enable innovators to create intellectual properties."
"We're build an AI assistant that helps tenants respond to emails with their landlords."
"This is your legal brain... on AI."
"In a world of global conflicts, where regulations become barriers to enter markets, Simplex helps retailers monitor regulation to help their products become and stay internationally compliant. Simplex, making market rules simple again."
Link: https://github.com/mathias-nillion/simplex-stanford
"Our project helps former inmates or people with minor and non-violent criminal history to find a job using a chat box that asks them questions about their criminal history, and then drafts letters of declaration that they or their lawyers can use to file an application to seal their criminal record."
Team: 2 lawyers, 3 engineers
"An LLM assistant to create FinTech dossiers for approval with the financial regulator."
"Compliance assistant for Wealth Management advisors."
"Annon PDF anonymizes PDF files by replacing PII info with fake replacements. To build AI tools for lawyers, real-life documents needs to be used either as training data for LLMs or in a RAG system. Our tool enables such tools to be built while ensuring client privacy."
"Finding related patents--done right."
"Train for your next deposition or trial. Train to Win."
Link: https://discord.com/channels/1105981880517525566/1226666009222447224/1226666009222447224
"Struggling with the implications of T&C in the fine print?
We do too! That's why we're building a solution to find out which fine print impacts you the most."
"Streamline discovery disputes in litigation, draft motions in minutes!"
"The first, the best, and the most accessible innovative "couch to lawsuit settlement" pipeline ever.
Forget the bus stop lawyer, give Suzie a call, and move into the 21st century courtroom from your literal couch.
Suzie always answers and wants to chat about your marriage troubles, slips, oopsie daisies, and work frustrations.
Ya know, the lawyer you can count on. Suzie will help you guide your story, from getting the facts lined up to asking the right questions, Suzie switches your mile high stress with mile high settlements."
"We are building an LLM that helps junior lawyer speed up reviewing documents in due diligence projects.Prepare documents and questions, and it gives you answers with basis location.
because llm doesn't ensure accuracy, it doesnt mean to replace the whole process of documents reviewing, but just speeds up the process of reviewing by providing information and basis location."
"You're ready to sign your residential rental agreement with a landlord, but you're wondering... is this weird clause legal? 🤔
Our tool will help you know if your contract contains illegal clauses! Just upload your PDF contract to check before signing it, and find out!"
Link: https://github.com/BinaryBrain/is-your-contract-legal
"Was your flight to/from the EU delayed or cancelled? Great - EU law says you’re entitled to compensation!
But wait - how can you actually claim it? Well, today you have 3 options:
#1 DIY (time-consuming)
#2 use one of many startups handling such claims (expensive)
#3 call a lawyer (even more expensive)
We’re building a cheap and efficient way to get you your money!"
"Based on Company documents, it creates a Due Diligence report draft.
It would also highlight all areas with possible red flags or uncertainties."
"We're building an AI assistant (AI-based Prenuptial Agreement Generator) that makes drafting prenuptial agreements faster and easier for couples. Prenuptial agreements can save couples significant money in case of unfortunate circumstances (accidents, medical debt).
However, many couples lack access to proper legal support. By leveraging RAG technology, our tool empowers couples to better understand the legal process."
"We're building a collection of hierarchical LLM Agents which can solve complex and general legal questions (providing legal information) based on primary source legislation. A collection or swarm of agents will be able to conduct smart legal ressearch by traversing source legislation with provided tools and actions."
"Opensource.lawyer is a devtool that helps choosing the right license for open-source projects.
every year, 62m+ open source projects are created on github. but most of us know only about the MIT license. we're building a tool to help open source authors to choose the right license, depending on their goals."
"We are building a Risk Factor Analysis Tool to help lawyers streamline the securities disclosure process and provide them with valuable insights."
"An application that creates due diligence reports of merger and acquisition agreements."
"Our project is a legal learning management platform on real-life cases to help young lawyers improve practice-ready skills and build confidence through real-life scenarios. Also, our platform can serve as a continuing learning education platform for seasoned lawyers."
"Advanced Document Organization and Semantic Interaction System (ADOSIS) organizes all the documents for a company or a legal case into virtual hierarchies.
It transforms a set of files to a self organized searchable platform, tagging the files and adding QA on the files in the interface."
"Our project leverages LLMs to help early-stage startups generate invention disclosures and kick-off their provisional patent filing process."
"A copilot agent that assists you by providing answers from the text, focusing on finding specific information rather than interpreting the document's overall meaning.
What we do: We train junior law firm associates with critical skills such as negotiation, communication, and redlining, ensuring their value and resilience in the evolving legal landscape. Our product offers a cutting-edge training platform that provides immersive simulated negotiation scenarios. The simulated negotiation will be through a Word document and in redline, and the AI will act as a law firm partner who is reviewing a junior's work."
"A tool for legal professionals that automates post-merger tasks.. and more."
"LLM are bad at problem that need planning and reasoning. With legal document processing increasing company hire junior attorney or use AI wrapper to get first good draft of contract or cases. But, after every draft retrieval from contracts SENIOR ATTORNEY still need to REVIEW these first draft causing loss of time and money.
This hackathon we are augmented our work in knowledge graphs based LLM to build a real senior attorney assistance called "Mike Ross". Here, we show as how a senior agent can use AI to make it learn and use it to automate first review of a junior attorney draft.
During the hackathon, we learned that one core deliverables will be confidence of each value retrived from AI and mark low confidence values only for review in AI. But, finding the confidence from LLM is not trivial - we iterated with prompts - created our criteria for evaluation - Used logprob and ROUGE like eval to improve accuracy of confidence. And finally reached a happy spot ...."
Mahesh Yadav
"PointOne (https://pointone.ai/) automates timekeeping for law firms.
We use AI to figure out what work a lawyer is doing on their computer and automatically generate their time entries for them.
For our project today, we’re adding a feature that retroactively generates a list of time entries based on a user’s emails and calendar events over a given time period (eg. last week)."
Adrian Parlow
"The idea started with the name of Immigration Solver at the 1st Stanford LLM hackathon.
What was once a simple letter of recommendation generator platform for O1, EB1A and EB2 NIW visas, is now a leading AI powered marketplace for immigrants in the United States.
The platform is used by 1000+ immigrants and 10+ lawyers to help draft O1, EB1A and EB2 NIW petitions with the help of AI and legal experts.Based on the recent success, we continue to expand in adjacent markets such as Canada.
Our goal is to help immigrants gain freedom and support them in their dreams."
Link: https://opensphere.ai/
Get your free evaluation for U.S visas: https://opensphere.ai/best-visa-for-you
Connect: https://www.linkedin.com/in/atalovesyou
"An Ai that notifies you if you have received a parking ticket, helps you figure out if and how you can contest it.
How it Works: Enter your license plate number. You will get a push notification if you recive a ticket and Ai will give you tips on how you can contest it based on the specific parking violation."
IPWiz tackles major IP challenges like complex intake processes, mitigate claim challenges, patent defendability concerns, changing laws, and high costs. Its solution emphasizes clarity, patent security, collaborative features, and cost-effectiveness, simplifying and streamlining IP management.
Link: https://IPWiz.tech
"We are building, Lawgic Legal Track employs AI to streamline legal workflows, automating administrative tasks and ensuring complete transparency and efficiency in client-lawyer interactions."
Link: https://legaltrack.ai/
"We are building a genAI-powered contract drafting and analysis word plugin for lawyers in the financial services space. Specifically, we are targeting lawyers who work with transactional documents in commercial real estate, M&A, and at investment funds."
"An LLM to register trademarks".
https://www.juridicon.ai/trademark-assistant-gai
I am Carlos Valderrma from Legal Paradox.
"AI-TC revolutionizes real estate transactions by offering an AI-powered assistant that simplifies document comprehension for buyers and enhances operational efficiency for agents, saving both time and money."
"Patents Wiki is an AI powered patent compare database for everyone. Explore prior art, invention, specifications, and claims from one powerful product designed for the global litigator and inventor."
The Hackathon was followed by the Technology & Applications of Computational Contracts in Insurance & Finance day and a half workshop as a part of the longstanding CodeX Insurance Initiative.
Panel Discussions were under Chatham House Rules so the details cannot be published.
"A computable contact is one that provides a computable specification for the terms and conditions of the contract. Given a computable contract and a set of circumstances, it should be possible for one to determine compliance of those circumstances with the contract’s terms and conditions in a purely mechanical way – without the help of human experts and without further clarification from the contracting parties."
Source: Extracted from a CodeX blog post that provided a transcript of a presentation on formal contract languages at the Computable Contracts Session of the 2021 FutureLaw Conference.
One of the highlights from the workshop was a presentation of the CodeX Insurance Analyst (a consumer facing prototype) addressing an insurance needs analysis use case by CodeX Computable Contracts Research Developer Preston Carlson.
The scenario presented in the demo is a common one, that if you have ever rented a car, you will have experienced.
Consumers are routinely presented with the option to take up rental car damage insurance coverage when they pick up their rental car.
It is currently very difficult for the consumer to understand what coverage they already have to determine whether or not they are already covered or will be covered by paying using a specific credit or debit card, as an example.
The required analysis can get complicated very quickly, and becomes almost impossible for any consumer to perform "in the spur of the moment".
In an abundance of caution, the majority of consumers opt to take up the rental company coverage.
If the consumer had the benefit of being fully informed (either by an insurance broker, lawyer or a technical application such as the CodeX Insurance Analyst) they might have discovered that they already had coverage via one of their other existing insurance policies.
Start by considering the following:
1️⃣ Private car insurance coverage might extend to cover rentals as well;
2️⃣ Travel insurance might indirectly cover rental car damage (or maybe not);
3️⃣ Insurance coverage might automatically apply if the consumer decides to pay the rental fees using their credit or debit card; or
4️⃣ Even if the consumer is already covered, the rental company car damage insurance coverage might offer a unique advantage, such as a zero deductible excess, that is only available via the rental car damage insurance policy.
The CodeX Insurance Analyst (CIA) application is designed with an eye to minimize insurance costs whilst ensuring no gaps in coverage.
It provides transparency, as the application provides immediate feedback across multiple insurance policies.
To attempt to achieve the same result manually is not easy, and consumers don't have the legal skills or time to manually review all of their insurance policies.
You can try the CodeX Insurance Analyst (CIA) application for yourself here at:
https://insurance.stanford.edu.
Inspired by the CodeX Insurance Analyst (CIA) developed by CodeX Computable Contracts Research Developer Preston Carlson, our innovative Rental Car Damage Insurance Coverage ➲ Legal Wizard includes Embedded Lawyer-Logic that is designed to help you make a fully-informed decision whether to opt to take up the Rental Car provider's Collision Damage Waiver?
* Note: As is our established practice we have published this Legal Wizard to our website whilst we are only in the initial stages of its development.
Please check the link regularly, and let us know if this Legal Wizard is of interest to you so that we can gauge the level or market demand and obtain you requests for features as we continue to iterate and improve the Legal Wizard.
In order to better understand consumer experiences with Rental Car Insurance, I searched for some additional context and found a relevant blog article titles "The Truth about Rental Car Insurance You Need to Know" by Ed Perkins dated 20 September 2023.
"Consumer advocate Ed Perkins, the founding editor of the Consumer Reports Travel Letter, has been writing about travel for more than three decades."
The introduction to the blog article is extracted below:
"No matter how often I cover the problems that come with collision damage under car rental insurance, consumers continue to share their frustrations.
A reader recently emailed me:
'I'm interested in finding a complete bumper-to-bumper, 'drop off the keys at the counter' car rental coverage when I travel. I recently thought I had complete coverage though a credit card, but when my car suffered a small dent, the car rental company flat-out refused to document that it actually lost income from having the car rental in the shop (which took three weeks to repair a small dent). I was left with responsibility.'
This experience is uncommon, but not unique.
Car rental companies desperately want you to buy their collision damage waiver, or CDW (sometimes also known as loss damage waiver, or LDW), and will go to great lengths to make you pay for even the smallest damages if you don't.
Here's what you need to know before you shell out for car rental insurance on your next vacation ..."
The detailed "What you need to know" part of the article includes the following headings:
The 1-way to drop your keys and forget it!
CDW Insurance is ridiculously overpriced
You'll pay up-front for all damage
Damage is more than fixing a dent
You need General Liability Protection
Alternatives to Rental Company Collision Damage Waivers
1️⃣ Pay with your Existing Car Insurance;
2️⃣ Use Credit Card Benefits;
3️⃣ Third-Party Car Rental Insurance;
The Uncertainty Principle
Key Takeaway
"The takeaway from all this: No alternative source of collision damage coverage-your auto insurance, your credit card, or a third party-completely isolates you from risk. Lots of travelers successfully rely on lower-cost alternatives to CDW without encountering any problems, sure.
Occasionally, however, your own insurance, credit card, or third-party policy may not cover an unusual charge a rental company might impose.
If you're unwilling to accept any risk at all, you might want to buy that overpriced CDW. For most travelers and circumstances, credit card or third-party coverage is sufficient, and primary coverage is a lot better than secondary."
Inspired by Dr. Megan Ma's presentation on the use of AI to help generate a 1st draft for a logic program from an insurance policy, I created a similar AI prompt, but instead of supplying the insurance policy, I just asked the AI to use any insurance policy it could find.
AI Prompt:
Take any insurance policy, list exclusions verbatim and rewrite them as extensional database relations for logic programs....
AI Output:
Here are some exclusions from a sample cyber insurance policy, listed verbatim and rewritten as extensional database relations for logic programs:
Verbatim Exclusions:
"arising out of, based upon or attributable to any dishonest, fraudulent, criminal or malicious act, error or omission, or any intentional or knowing violation of the law, if committed by any: (1) Insured; (2) past or present partner, principal, director, officer, trustee, or employee of any Organization; or (3) agent of any Organization;"
"alleging, based upon, arising out of or attributable to any: (1) prior or pending litigation, claims, demands, arbitration, administrative or regulatory proceeding or investigation filed or commenced on or before the Prior and Pending Litigation Date; or (2) fact, circumstance, situation, transaction or event underlying or alleged in such litigation,"
"arising out of, based upon or attributable to any unsolicited dissemination of faxes, emails or other communications by or on behalf of an Insured Organization or Third Party Contractor, including: (1) actions brought under any federal, state or local statute, ordinance or regulation, including the Federal Telephone Consumer Protection Act of 1991; or (2) cost of complying with any injunctive relief or settlement agreements related to any of the foregoing;"
Rewritten as Extensional Database Relations:
"exclusion(X) :-
dishonest_act(X),
committed_by(X, insured).
exclusion(X) :-
dishonest_act(X),
committed_by(X, Y),
(partner(Y); principal(Y); director(Y); officer(Y); trustee(Y); employee(Y)).
exclusion(X) :-
dishonest_act(X),
committed_by(X, agent_of_organization).
exclusion(X) :-
prior_or_pending_litigation(X),
filed_before(X, prior_and_pending_date).
exclusion(X) :-
prior_or_pending_litigation(X),
underlying_fact(X, Y),
alleged_in(Y, prior_or_pending_litigation).
exclusion(X) :-
unsolicited_communication(X),
sent_by(X, insured).
exclusion(X) :-
unsolicited_communication(X),
sent_on_behalf(X, insured).
exclusion(X) :-
unsolicited_communication(X),
related_relief(X, injunctive_relief).
exclusion(X) :-
unsolicited_communication(X),
related_settlement(X, settlement_agreement)."
The conference itself was jam-packed with too many insights to adequately cover here.
It was great to be able to connect with so many tech savvy academics, legal tech startup founders & attorneys.
The following is a summary of some of the key takeaways, in no particulate order, I gained from attending.
As I have spare time I intend to further update this part of the blog article.
Please check back at a later date for my additional notes & takeaway insights.
The general consensus is that AI has a long way to go (and may not ever reach the point where) its output can be relied upon without the intervention or oversight of an experienced Attorney.
One example I regularly provide to articulate the current state of play, is a comparison between the speed and quality of any of the current AI offerings versus an automated expert-system with incorporating embedded "Lawyer-Logic".
When I run the simulation in my mind to determine which solution is capable of being the first to generate a final useable legal document, the expert-system always wins.
One of the downsides to designing & building legal expert-systems is that they take a long time, and consequently for lawyers who bill their time in 6-minute increments, a large $$ investment to build & test a final legal product.
What if we could harness the best of both approaches?
On the agenda for the conference, was the hybrid logic + AI approach which was described as gaining traction.
Harry Surden Professor of Law at University of Colorado Law School & Associate Director & one of the co-founders of CodeX, provided a Mental Model for thinking about AI as having a top 3rd year law student as your intern.
Amazing drafts can really be beneficial, but are not turn key, that is, not ready to file with the Court.
Also AI may be able to provide answers to easy cases, that is, cases that most lawyers would agree on, but out there in the public there is no knowledge or consistency about how these cases would fare in Court.
“The work SALI is doing to establish industry standards on matter types addresses two critical problems facing the legal industry: first, our continuing quest for value-orientation and second, extreme inefficiency in the buying and selling of legal services.”
Jae Um
“What is exciting about the SALI initiative is the ability to develop an underlying framework upon which to build smarter systems.”
Rebecca Benavides, Microsoft
“A modern clear legal standard provides the unifying language between firm and client, offering clarity and transparency while promoting efficiency and opening the door to accelerated innovation. SALI’s standard does that.”
Vincent J. Cordo, Jr., Holland & Knight
"This week we have Damien Riehl, VP, Litigation Workflow and Analytics Content at FastCase, and one of the drivers behind SALI (Standards Advancement for the Legal Industry.)"
Further reading: 3 Geeks & a Law Blog article "The Bullshitter, The Searcher, and The Researcher" – Damien Riehl on the Dynamic Shift in How the Legal Profession Will Leverage Standards and Artificial Intelligence by Greg Lambert & Marlene Gebauer on February 8, 2023.
Version 1 of SALI had around 1,500 tags, whereas as version 2 of SALU has increased the number to more than 10,000+ tags, and moved from the use of spreadsheets to the use of XML which facilitates easier adoption.
After the launch of version 2, SALI has seen unprecedented uptake and is increasingly being adopted by major law firms and software providers.
The best way to browse the SALI LMSS 2.0 standard is through the Stanford WebProtégé tool:
Select LMSS 2.0 to see the latest official release; to see the most recent working draft for comment, select DRAFT LMSS 2.x
SALI on GitHub
Excel version:
Dr Megan Ma presented about LLM's equalizing effects.
Part of Dr. Ma's work includes simulation of legal behavior using AI to provide legal training.
The demonstration was built around a scenario for M&A negotiations occurring before signing of the letter of intent.
Varying levels of lawyer experience could be selected, with additional layers of complexity added based on the level of lawyer experience selected.
The LLM provided negotiation responses from Opposing Robo-Counsel...
Tom Melling, CEO of Safari SOP presented on eService of Process.
The solution to the "Legal Junkyard" caused by existing laws that require registered agents and physical paper service at their registered address, is to have companies consent to only be served electronically at their website specifically designed fo this purpose as a Served Document Intake Portal.
This solution provides immediate acknowledgement of receipt.
Companies can specify that they only want subpoena's or will accept all docs electronically.
As the website could be served 24/7, the server will need to acknowledge that if they serve after a certain hour, they agree that it will treated as next day service.
The online eService process can wrap legal terms, capture payments, provide authentication, and enable bulk service.
Further reading: Safari SOP website
Below are my bullet point style notes take during David Engstrom's closing keynote addess where I have attempted to provide the gist of his key points.
TAR Wars
"Seed set" fights
- relevantce Rule 26(b)(1)
- work product Rule 26(b)(3)
Propel the haves further ahead...
Haves still have privileged access to data
Case Outcomes Data - Holy Grail
Most cases - settlements or cloaked by NDA's...
Walmart - first will legal data analytics
Lack of public access to law
75% at least 1 side lacks a lawyer..
Debt Collection 24%
Multiple case types 16%
Tort 7%
Other 24%
Mortgage foreclosure 11%
Landlord-tenant 19%
Lawyers' Monolpoly
Rule 5.4
Inevitable - Utah sandbox, Arizona...
impressive examples..
Reg Reform - movement
in CA, State Bar group to make a sandbox recommendation - shutdown by the judges...
New - the sea of junk... bad advice.
DIY direction - looks like legal advice and free... M. Hagan leading the way...
Prohibit AI from doing legal
Court Approved - service providers.. efiling...
Make (Stanford - LA Superior Court) or Buy?
A: court-linked AI tools
B: robust 3rd party "legal tech" ecosystem
Different tech infrastructures
Filing Fairness Project
7 states...
14 thousand local court jurisdictions - this defeats the scale...
Court technical capacity woefully thin...
Partner - only partial answer...
Path to a world where we try to fold LLM's into the existing system.. rethink what that system can and should look like.... ADR systems (not currently used).
While I missed the last day, I understand that one of the highlights was a demo of Norm.Ai from John Nay's LinkedIn post:
Then, on Friday, the Dr. Megan Ma led event, “Steering the Intersection: Moving from Exploration to Maturity on the Interplay of Generative AI and the Law,” at CodeX, The Stanford Center for Legal Informatics was a perfect collaborative venue to demo Norm Ai regulatory AI agents and discuss the different levels of LLM-powered AI agents that we are rolling out.
Social Sharing Image: Sourced from the CodeX FutureLaw 2024 Conference Home Page.
Credits: This blog article was written by James D. Ford Esq., GAICD CIPP/US CC | Attorney-at-Law, Blue Ocean Law Group℠.
State of California Bar Number: 346590
Important Notice:
This blog article is intended for general interest + information only.
It is not legal advice, nor should it be relied upon or used as such.
We recommend you always consult a lawyer for legal advice specifically tailored to your needs & circumstances.
Your comment has been received and we will approve it shortly.