Lemkin AI

Peace is perceived to be passive. While defense and security is always on the cutting-edge of innovations, peace is considered obscure and nebulous. But, let’s create and defend peace. Rather than a military industrial complex, let’s build out a peace infrastructure that helps the process of peacebuilding for sustainable peace. The main problems we’ve identified are:

1. Barriers to begin peace talks
2. Ebbs and flows of International Community’s interest dictating what conflicts are resolved
3. Peace-seeking parties in conflicts but without the tools and support to make peace
4. Critical information access for citizens during conflicts
5. Citizen participation and education in transition to new society

And the quantity of these conflicts may only increase going forward. Peace is a pre-requisite to prosperity, but there’s little attempts at innovation and scaling around peacebuilding.

With our pilot program in Myanmar in particular, we are working with a long-standing conflict around ethnic minorities and an oppressive military. The myriad of cultures, ethnicities, and visions makes it an involuted terrain for anybody wanting to help broker peace. Many have dedicated their lives to helping Myanmar, but on the whole there has been waning interest in Myanmar’s plight for democracy and an end to its violence.

The impact spans generations as they are lost to the conflicts. Proper institutions are fragile and fraught with discrimination, leaving the education system unable to serve the students.

Much of this has to do with the larger, international power imbalances. We can take it all the way back to colonialism, but what has been constant throughout is the heavy reliance on Western powers, despite their caprice. Their geopolitical calculations precede real, on-the-ground needs, leaving the population of Myanmar isolated and at times abandoned.

This is Myanmar’s specific conflict, but the elements may not be unique to Myanmar. As the world shifts to a more multipolar world, theory suggests that more smaller conflicts will arise. It is therefore crucial for there to be solutions that, irrespective of Western interest or disinterest, empower peace-seeking parties to have agency over their lives.

Solution
The solution is a suite of AI-powered software applications that augment the peace negotiations and peacebuilding process.

Lemkin AI is a software environment that helps users quickly synthesize information, identify sentiments, values, visions, and objectives, and create action plans for negotiations. Lemkin AI functions as a “second-chair” for each peace-seeking agent (and for their real legal representation) while also function as a mediator for the whole peace process.

Arbiter is a chatbot designed for citizens. It provides real-time conflict updates and dispels misinformation during conflicts. Post-conflict, it serves as a 24/7 help desk for citizens to inquire about their new society, rights, and responsibilities in a transitioning democracy. The chatbot is available on various social media messaging platforms like WhatsApp and Facebook Messenger.

How it works
Lemkin AI’s workflow for users is simple. Peace-seeking parties will upload their documents (of any nature: declarations of independence, policy memos, list of grievances, etc.) to the software. In the backend Lemkin uses AI to synthesize that information into values, visions/interests, and objectives. Lemkin also uses sentiment analysis to assign each of these items with a value function, which it then uses to create a ranked order list for each participating peace-seeking party.

The ranked order list is used to create an action plan for negotiations. But, each party is also given the ability to see the ranked order list and the conflicts that exist between parties to adjust their value functions in pursuit of a peaceful outcome.

Arbiter uses RAG (retrieval augmented generation), a technique to have a central library of articles and updates on the conflict to pull from. Arbiter uses only this central library to respond to citizen questions in its chatbot. This central library will be open for anyone to access and see to ensure transparency and for all items to be vetted.

Technology
The technology underlying both Lemkin AI and Arbiter is the nascent Large–Language Model (LLM) technology for generative AI combined with RAG. This technology is made accessible through user friendly interfaces.

Benefits
This suite of solutions makes the process of peace negotiations and peacebuilding more accessible, more inclusive, and more efficient.

Accessibility
For Lemkin AI, the prompts in the backend ensure that the uploaded documents don’t require any special, expert legalese. Any format and eventually language is accepted for Lemkin AI to distill and begin its work.

Inclusivity
Rather than being limited by the capacity of personnel, Lemkin AI can work on a myriad of documents that may take a whole team otherwise. This allows more parties to join the peace process.

For Arbiter, as a society transitions post-conflict, all citizens can take advantage of their newfound freedoms, liberties, and rights. Arbiter ensures that citizens are educated on these rights, right from their fingertips.

Efficiency
Lastly, it is made more efficient because it is a matter of computing power rather than IC interest and personnel they send. LLMs can synthesize faster and they are only improving in speed and quality.

Pilot: Myanmar population and diaspora

For the pilot of the Lemkin software solution, we are focusing on the population of Myanmar. The country has been oppressed by a brutal military coup since 2021, which awakened the whole of the population to the military’s ruthless, indiscriminate appetite for violence and power, not just the minority ethnic groups who have long been victimized by the military.

Many minority ethnic groups, who have been fighting the military for decades, are still pursuing self-sovereignty, even independence. Therefore, it is crucial for all stakeholders of Myanmar to come together, not just against their common enemy in the military, but to plan for what comes afterward. 

The involuted nature of Myanmar’s conflicts is fraught with memories and scars from social, political, religious, and ethnic tensions. Lemkin hopes to facilitate these talks and the designs of a new Myanmar—in whatever form the parties agree on. Myanmar has already lost generations to wars, violence, and genocide. Lemkin is a tool that empowers Myanmar to pursue their future of peace and prosperity on their terms.

The pieces are all there, ready to be put into play with the shadow government in the form of the National Unity Government (NUG). It is not yet fully representative, but is the most inclusive yet and has aims to be legitimate.

Arbiter, a part of the Lemkin suite, is a chatbot that gives citizens easy-access to crucial information. During conflicts, Arbiter serves citizens up-to-date, vetted information. Post-conflict, as the society is in transition, Arbiter serves citizens as an advocate, giving them accurate information on their new rights and responsibilities. This empowers each citizen to live fully in their new, freer society. This chatbot is easily accessible in their choice of familiar messaging interfaces (eg, WhatsApp, Facebook Messenger, etc). 

There is a heightened level of cohesion around Myanmar that must be made use of for positive ends.  The long promised future of a liberalized Myanmar cannot be achieved if we once again have Myanmar slip into its web of conflicts. Instead, we are given a prime opportunity to see Myanmar create a peaceful country of their design and prosperous economy of their accord. 

An intervention with Lemkin can help Myanmar, not just overcome their current oppressors of the coup, but also end one of the world’s longest civil wars. Countless lives can be spared, and countless futures spurred to live up to their potential.

Scale
Expanding from Myanmar, we see Lemkin being useful in other current and future conflicts. As the world continues to become ever-more connected and as some predict a transition to a multipolar world order, the number of small conflicts will, in theory, increase. The IC, however, will continue in its caprice and limited capacity. Giving those peace-seeking parties in conflict the tools to produce peace gives true agency. And for lawyers—whom we see as still vital figures is peacebuilding—Lemkin frees up their time to better serve their clients, efficiently and effectively.  

Given the interdisciplinary nature of this project, it requires a wide-range of experts. As a solo-team for now, I sit at the intersection of international development, humanitarian aid, technology, and language-skills. During my time in the Thai-Burma border region, I experienced the challenges faced by refugees in conflict zones. As a university lecturer, I created interactive classrooms that encouraged open discussion and critical thinking. Despite initial discomfort, my students embraced the opportunity to debate and engage with sensitive topics, leading to collaborative projects beyond the classroom.

Here, I learned to take a considered approach to syllabus design and lesson plans. Borrowing from anthropology, ethnography, and design research frameworks, I implemented my own student-centric user experience study to design a new course. These skills are essential to both designing the Lemkin software, but also for interfacing with the users of Lemkin, who are in sensitive, precarious situations.

A user-centric approach puts users at the center and gives space to their needs. This is especially important in the context of Lemkin for 2 main reasons. First, conflicts can take place anywhere in the world, in any culture or language. The design must be able to accommodate this diversity. Secondly, these cultural or language differences may be a part—if not the source—of the conflict. Lemkin, in its front-facing designs and backend engineering of prompts and code must be sensitive to this.

Coupling this with my experience and connections from working at a pro bono law firm that focuses on peace negotiations and post-conflict building will allow for tailored, expert input to the needs of the peacebuilding process. From the essential roles of the lawyers and proven negotiation frameworks to bottlenecks and barriers to the process, expert experience is important to bring in early.

A large part of Lemkin’s AI capabilities is in prompting (or prompt engineering) the AI to elicit a desired output. This is a nascent skill, born from the rise of generative AI, but important to this is a broad understanding of the underlying technology and language skills. My training in clear writing and critical thinking come from my undergraduate studies in English and Master of Liberal Arts degree. Additionally, I have been selected to join the most recent cohort of Stanford’s Code in Place course that is taken from their intro computer science course CS106A. I hope to use this as a starting point to better understand the underlying code and research around AI.

I have also bootstrapped a side project, which has given me a vast amount of learnings around design, user research, working with developers, and data cleaning and management.

Sitting at this unique intersection will be important as the founder of this product. The most important knowledge from this, however, is to know where in the product pipeline expert input is needed. Lemkin comes from a place of empathy and humility, and as a product and a team, will continue to be driven by these values.

Lemkin AI is currently in the prototype phase of development. We have rigorously tested prompts and LLM variations to find what yields the best results. This is the core of the product. Lemkin AI also has a ranked choice order component and this function has been developed and receives its input from Lemkin’s prompted value functions.

We additionally have the preliminary layout and design of the software, but are yet to develop it into a refined user experience. The backend of the solution is complete (though quality testing will continue) and we are working on the frontend.

We believe that Lemkin can be a force for good that addresses the direct needs of global conflicts while also shifting the perspective around the field of peacebuilding. In pursuing these ambitious goals we need equally ambitious partners and we believe we are aligned with MIT Solve in these values and goals.

Specifically, we’re looking build out the software from prototype to a working product as quickly as possible to implement in the Myanmar context. At our current capacity, we are not able to build or fund the build of the software in a timely manner. We hope that through Solve we can receive the validation, financial backing, and connections to meet this objective.

Given that we are still a prototype while also looking to be setup as a not-for-profit, venture funding is a difficult path. A for-profit setup was considered (and is not completely out of the question), but the use case of Lemkin in peace negotiations requires strict adherence to unambiguously objective standards, which a non-profit may be better at upholding. Solve gives Lemkin the opportunity to build a fully capable product from a prototype while also establishing an appropriate and stable non-profit structure to support it.

With this substantive reason for applying to Solve, we also hope to change the narrative around the field of peacebuilding. Peace is too frequently seen as passive and simply the absence of violence. But, we see peace as something that is worked at, something intentional, considered, designed, and sustained. But, while there are tools (and an entire industry) around defense and security, peace is neglected. A part of this is that peacebuilding requires a lot of soft skills and interactions between people. We are not trying to forego or replace this. In fact, we’re hope to enhance those interactions and facilitate the negotiations and mediations. Augmenting these processes with technology opens the individuals involved to create more personable and intimate relations with each other through more time spent on the qualitative nuances rather than menial tasks. 

Our grand vision is not only for Lemkin to be a peacebuilding tool, but for an entire industry to form around peacebuilding to shift from a military-industrial complex to a peace-industrial complex. And from peace to prosperity, we hope to identify a new vertical in establishing a clear pipeline that leads to sustained peace and inclusive prosperity.

The Lemkin AI solution looks to leverage the nascent technology of generative AI to augment the peace negotiations and peacebuilding process. Generative AI allows for quick, comprehensive synthesis of mounds of information that would typically take extensive manpower, man hours, and expertise. Lemkin’s innovation lies in its application of generative AI and its user experience, designed to lower peacebuilding barriers to entry for a more inclusive, more accessible solution.

The innovation is seen in how it fuses proven peace negotiations strategies and frameworks with AI technology solutions in high leverage points of the workflow. For example, the Public International Law & Policy Group (PILPG), a pro bono law firm that specializes in peace negotiations and peacebuilding, has identified the essential value-add role of a lawyer involved in peace negotiations. PILPG has said that in the early stages, synthesizing the myriad of information for clients then creating an action plan for the clients from it is a crucial part of a lawyer’s job.

Lemkin AI is capable of doing this same work with less manpower and hours. Ideally, this is used by lawyers in representing their clients who are the peace-seeking agents. However, Lemkin AI has made this part of the process more inclusive with the use of AI, allowing peace-seeking agents without the resources or connections to have legal representation to begin the process of peace negotiations. Formal legal language or format is not required (and we hope to expand the capabilities to include many more languages). This significantly lowers the barriers to entry, smoothening much of the friction involved with beginning the peace negotiations when the parties are ready.

Much of the AI features are embedded in the background, making the experience seamless. The prompts that trigger a response and output from the AI are created (and revised) by the Lemkin team through iterative testing and evaluation. In the backend multiple prompts are chained together to elicit more detailed or tailored responses, which are then displayed to the user in an easily digestible format. The user interface that is created with the prompts in the backend thereby requires no technical understanding of AI prompting or particular language skills by the user.

Another Lemkin AI innovation is the use of value functions to rank choices for negotiation, forming an action plan. Quantitative values are assigned through sentiment analysis and can be adjusted by each party. This process establishes a negotiation agenda by prioritizing issues. Each party receives a text to modify based on their perspectives, inspired by the single-text negotiation strategy from the Egypt-Israel conflict at the Camp David Accords.

While implementing nascent technology, Lemkin AI is built on proven principles such as this. The innovative blend gives Lemkin AI its value to help lead conflicts to peace.

The Arbiter chatbot is innovative in its application. Using similar technology as Lemkin AI, this chatbot connects citizens with a vetted and transparent corpus of news articles and policy documents through familiar apps such as WhatsApp and Facebook. 

To begin with the problem, the barrier to entry to peace negotiations and peacebuilding is too high. Conflicts may have peace-seeking parties, but these parties may be lacking in resources to begin the process. All the while, the International Community may not have the capacity to address that specific conflict. This only prolongs conflicts, increasing the number of lives lost and the lives put on hold.

Lemkin’s solution significantly lowers the barrier to entry by giving peace-seeking parties a tool that functions as a lawyer representing their interests as well as a mediator between the parties involved in a negotiation.

Ultimately, impact on this problem is measured in the peace it produces. Lemkin is able to have a positive impact on these enumerated problem because it is based on peace negotiation frameworks and perspectives of experts in the field. Rather than simply pasting technology to the process haphazardly, Lemkin is looking to create a workflow based on proven peace negotiation strategies, augmented by technology.

The second problem is the access to reliable information during conflicts and post-conflict around updates, warnings, and rights. With the prevalence of social media and the central place social media plays in many conflict-affected regions, information may be aplenty, but reliable information may be obfuscated or lost in the noise. And when a society is transitioning post-conflict to a new society with new rights and new responsibilities, old habits may die hard. That is the case for citizens, but also for potential bad actors accustomed to getting away with abuses of power. Citizens could do with an easily accessible rights and responsibilities handbook of sorts to reference. There may be new democratic activities—eg, registering to vote—or social welfare programs one may be eligible for. Additionally, there may be individual actors who abuse their power. Citizens may be accustomed to accepting it for fear of repercussions. But, knowing their new rights in a new society should empower them to hold those in power to account.

Arbiter, the second application in the Lemkin suite, is a chatbot that is accessible through already familiar means. This gives citizens easy, 24/7 access to such a rights and responsibilities handbook. And, given the chat interface, it isn’t a daunting tome, but a friendly, informal conversation.

In Myanmar, Facebook and the Internet are synonymous. Facebook Messenger is a natural way of communicating and already a part of the society’s tech culture. Through Facebook they find new friends, keep in touch with old friends scattered through the diaspora, travel the world through pictures uploaded by their diaspora friends or visitors from abroad, and even begin—and end—romantic relationships.

Giving citizens access to a chatbot that can help them understand their rights or help them verify if a rumor on Facebook is true wouldn’t be a stretch by any means.

Our goals:

• To make peacebuilding and peace negotiations more accessible and inclusive
• To empower peace-seeking agents in conflicts to design their peace without heavy reliance on IC
• To design a more robust system of post-conflict transitions to make peace sustainable
• To develop an industry around peacebuilding

Our success will be measured by:

• Shorter violent conflicts and quicker transition from conflict to beginning negotiations and coming to agreements
• Global metrics we’ll look at:
  • decreased death toll from conflict (including total, combat, civilian due to fighting)
  • decrease in number of organized armed groups
  • increased trust in government/state
  • decrease in human rights violations
  • Increased VC and philanthropic capital toward peacebuilding-specific software, tools, and entities
• Use of term “peace-seeking agent/party”

The past few years has seen a proliferation of generative AI applications and it is only increasing. On the more serious, corporate end of the spectrum, we see applications already in law, business and productivity softwares, and healthcare. Many early iterations of interacting with AI systems were in the form of chatbots, popularized by OpenAI’s ChatGPT. Lemkin AI puts much of the AI interactions in the backend, giving users a more seamless workflow. Lemkin will have prompts that trigger the AI to generate an output embedded into the software. All users have to do is upload their respective files and Lemkin will automatically process the files and generate outputs specific to the task of peace negotiations. We’ve created several specific buckets: Summaries, Values, Visions/Goals, and Objectives.

Lemkin has specific prompts for each of these buckets, eliciting relevant, quality outputs that Lemkin displays for users in a user-friendly design. In the backend we are using a prompting technique called prompt chaining. This method allows us to build on individual prompts to produce more detailed, relevant, and quality outputs for peace-seeking agents to use.

Lemkin AI uses RAG (retrieval augmented generation) technique to ensure that the prompts and outputs reference only the most relevant documents: the documents uploaded by the user. RAG allows Lemkin to remain a focused software for users.

The frontend user interface that users will interact with borrows from dashboard designs and professional software, like photo/video editing tools and software IDEs. This reflects our belief that peacebuilding is an industry of professionals and requires pro-tools. However, we also want to empower any peace-seeking agent to use this tool, and as such, we are working to make the design and functionality intuitive.

For the Arbiter citizen chatbot, we are similarly using RAG. Only, the uploaded documents will be centralized to the Lemkin team with open access to the public for transparency and vetting purposes. The chatbot that the citizens interact with will respond with only the information provided to them through the RAG technique, and there will be safeguards put in place to ensure Arbiter doesn’t give false information and rather states that it doesn’t know. The interface for this will be through existing social media messaging apps, such as WhatsApp and Facebook Messenger.

1

5 months formally

The nature of the work by nature necessitates diversity and inclusivity while also promoting it. On the substantive side of this coin, using the Lemkin software in the peacebuilding process opens the process up to more parties than traditional methods. Without Lemkin or tech-augmented processes, the peacebuilding is limited by the capacity of the personnel involved. More parties would then only slow or complicate the process. Using tech allows for the processes to have the capacity to be more inclusive and representative. Computing power is the only limitation now.

For example, if there are two main parties in conflict with each other, say Party A and Party B, their coming to the table does not require too much capacity. But, if Party A and Party B want to be truly representative of their constituents, they would require a lot more capacity to synthesize the various needs of their respective constituents to then go on to negotiate. In real world terms, there are civil society organizations and activists groups that represent a particular portion of the population. Ideally, having their voices heard and included in designing a new society would yield the more representative results. The manpower and expertise required to synthesize and distill may be overbearing during peace talks. But, with Lemkin’s AI solution, this can be done more quickly and eventually even with a higher quality output. This opens up the peacebuilding process to be more inclusive and thereby representative.

We are looking for ways to carry this spirit of diversity and inclusivity into the day to day of Lemkin. Even before we get into forming our team, we are already looking for ways to be more diverse and inclusive. We are intentional in our outreach for partners and are looking to bring in a wide range of feedback for the development of Lemkin.

With the pilot in Myanmar, we looking to include a diversity of voices in its implementation. We are currently talking with established community leaders as well as students and youth movements, bringing in voices across age groups, sex, region, religion, language, ethnic group. Ultimately, Lemkin as a product and an organization looks to go beyond giving marginalized populations just a voice, but giving them an actual say.

Lemkin currently considers a non-profit status makes the most sense. However, having a software as a central product may require upkeep and technical knowledge. Additionally, Lemkin may also provide services in peacebuilding contexts to operate the software in specific conflict contexts. We are therefore looking to have a model that blends reimbursement for services, grant funding, and beneficiary (alumni) support. 

The first 2 are straightforward. We may consider having a fee for our services and software access, but not for profit only enough for maintenance and overhead. Grant funding from large grantmaking institutions will be crucial partnerships to make. The third option, I believe may be the most interesting. We are exploring options around revenue-sharing in the new society that comes from the peace negotiation where Lemkin was used. This can be an innovative way of securing funding for subsequent conflicts while also serving as a mechanism for aligning interests. This is still in development.

Our current plan to become financially sustainable to is quickly run the Myanmar pilot to exhibit its usefulness while also making any necessary iterations to the product. The success in use for peace negotiations and peacebuilding will be most crucial.

As for the tech, though nascent, this is a vetted blend of AI techniques and user interfaces. In the for-profit world, we are seeing a larger interest in the application of AI rather than just in the foundational models. Investments are shifting to more industry specific products, rather than general AI, and from ‘AI companies’ to more AI embedded products.

Sequoia Capital, in their article on the second phase of generative AI has identified RAG as an important part of the development stack around emerging user interface paradigms for generative AI. It states that RAG is important in bringing in context, reducing hallucinations, and increasing truthfulness. The article also mentions new means of interfacing with generative AI. And in an interview with 20VC, Sam Altman of OpenAI stated that products that build with the assumption that foundational models will only get better will thrive.

With this, we see a new crop of philanthropy and giving. The efforts around data driven altruism coupled with ideas around existential risk may be re-focused to making positive impact in conflict zones that are presently on-going. Additionally, research initiatives around technology and institutions, like AI for Institutions, have identified LLM-augmented negotiations as point of potential impact.

The desire to support this technology and this mission is there. Lemkin must now prove itself useful in practice.