Intuit

Success means making sure high-level tech isn't limited to "Big Tech" hubs like Silicon Valley or San Diego. A successful dialogue would result in a real plan to share technical knowledge globally. The goal is to ensure any engineer, regardless of where they live, can build software with the same quality and rigor as those at the world's biggest companies. True success also means helping nations shift from being users of AI to architects of AI. We win when developers in emerging markets have the same access to tools, infrastructure, and mentorship as those in major corporations. Ultimately, the dialogue should focus on inviting everyone to participate in building the future, rather than just focusing on restrictions.

• AI capacity-building
• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Open-source software, open data and open AI models

These priorities aim to transform global talent into active participants in progress. AI capacity-building is vital because programs like AddisCoder, an intensive program that teaches high-level algorithms and programming to talented students in Ethiopia, prove that when you combine raw talent with high engineering standards, you create world-class innovators. To support this talent, we must prioritize open-source software and data. These serve as a "global public library," ensuring that the building blocks of AI are a shared resource rather than a locked gate. Additionally, focusing on linguistic and cultural implications ensures that AI development respects the diversity of the human experience rather than ignoring it. Finally, safety and trustworthiness are the essential foundations of any distributed system; without them, no technological advancement can be truly sustainable.

Training talent is a crucial first step, but it cannot stand alone. Without access to high-performance computing and local data centers, emerging ecosystems will continue to lose their best engineers to regions like the US or Europe. For AI development to be globally balanced, governance must ensure that engineers have the physical resources they need to innovate within their own communities.

In my experience, the biggest gap in planning is not talent but access to the tools needed to build. Through programs like AddisCoder, we have shown that strong engineering talent exists in many parts of the world. The real issue is that many engineers lack what I call compute sovereignty, meaning access to powerful computers, servers, and infrastructure needed to build advanced systems in their own countries. Without this access, even skilled engineers are often limited to building small applications on top of systems controlled elsewhere, usually in the US or Western Europe. This pushes many of them to leave their home countries in search of better resources and opportunities, which leads to brain drain and slows innovation locally. At the same time, open source technology creates a real opportunity. If the global community continues to support open software and data, local engineers can build tools that match their own languages, needs, and contexts without depending on large foreign tech companies. The goal should not only be to train more engineers. It should also be to give them the ability to build and scale where they are. This requires investment in real infrastructure such as computing power, servers, and data centers. The next step is not just using technology built elsewhere, but building it ourselves.

The AI Dialogue should bridge the gap between policy and actual engineering. Right now, global cooperation is stuck on ethics, but as a Senior Software Engineer, I see that the real barrier is technical access. The Dialogue can move the world toward a system where sharing computing power and engineering standards is the priority. Its most important role is making sure AI doesn't become a "closed club." By setting global standards for infrastructure, the Dialogue can ensure that talent in every country has the tools to build their own systems rather than just buying them from others. It should focus on keeping the "building blocks" of AI, like open-source models and clean data, free and available to everyone.

The Dialogue should build on talent first models like AddisCoder. That program showed that when students are given the same high level algorithms and engineering standards used at top universities, they are fully capable of performing at a world class level. The next step is to apply this same idea to AI. Instead of focusing only on basic training, we should be sharing the deeper architectural rigor required to design and build complex AI systems from the ground up. The real value of the AI Dialogue is shifting from education to execution. Programs like AddisCoder help create strong talent, but the Dialogue should go further by helping provide the hardware and infrastructure that allows those graduates to actually stay and work in their own countries. By connecting with open source ecosystems, the Dialogue can help ensure that engineers anywhere in the world have access to the same models, tools, and compute resources as those in major tech hubs. The goal is not just learning in isolation, but equal access to the building blocks of modern AI development. Its unique role is to prevent the pattern where talented engineers graduate and then leave because they cannot build at home. Instead, it should make it possible for them to stay, work on real systems, and help grow a local AI industry where they are.

In order to foster such a discussion, stakeholders need to participate in it through the lens of bridging the gap between technicality and policymaking. Engineers and architects can establish concrete metrics for achieving necessary standards of safety in systems, whereas academic instructors from AddisCoder-style programs can devise the appropriate methods of training new talents to reach the level of expertise. In addition to software, which the private sector often offers, compute grants that will give researchers from emerging markets the means to build are also crucial. The structure of the dialogue should focus on technical workshops and working groups that will discuss region-specific requirements and needs.

The most unrepresented perspective is represented by engineers and architects who are developing solutions within their own technology ecosystem. At the moment, conversations around AI governance are being led mostly by government leaders and representatives of established companies, overlooking the people who develop software in less developed markets. To include them in such discussions, we should create direct communication lines with their governance board. Other underrepresented perspectives are experts in low-resource languages and developers of open-source technologies that the world relies on today.

To foster meaningful dialogue, we can replace standard panel discussions with hackathons where participants would be challenged to develop a solution to a particular technical or ethical problem in a limited period of time. Such an approach will allow the attendees to stop making generalized statements and actually engage in developing real-world ideas. As another option, reverse mentoring, where engineers from resource-constrained countries will speak about how they innovate and avoid certain technological generations, may be useful. Lastly, the dialogue itself needs to involve sandboxes that will allow testing of AI policies in practice.

Instead of every country struggling to build its own massive data centers, regions can pool their resources to provide the high-performance hardware that local engineers and researchers need. This is a practical way to close the infrastructure gap, ensuring that talent developed through programs like AddisCoder has the actual power to build and innovate at home instead of moving abroad.