A Big Tech Partnership Playbook

Introduction

This case study looks at how the Icelandic government funded initiative, Almannarómur, was able to engage with Big Tech and build effective partnerships. We look at their approach and lessons learned and distill them into a playbook for partnership building.

Almannarómur is a centre for language technology responsible for implementing the Icelandic Language Technology Programme in accordance with a 2018 agreement with the Ministry of Education, Science and Culture. The overall goal of the Icelandic Language Technology Programme is ensuring the viability of the Icelandic language in the digital age.

Almannarómur’s mandate included ensuring the execution of scheduled projects, coordinating with industry stakeholders, and fostering communication between local and international partners. Almannarómur continues to be responsible for project management, including outsourcing core projects, managing the allocation of utilization grants to encourage the adoption of Icelandic language technology by businesses, and taking on promotional and advisory roles to advance Icelandic language technology within the domestic economy and among international tech companies. 

The work Almannarómur has carried out with notable success reveals much about the dynamics of collaboration between big tech companies and smaller organisations, diverse stakeholders, and even governments. While the goal in this case is to ensure the survival and preservation of Icelandic language and culture, their methods provide a broadly-relevant playbook that other organizations can use when engaging and building partnerships with the tech sector.

Almannarómur and OpenAI

Icelandic, a small language community of less than 400,000 speakers, was the first language outside of English to be included in the development phase of GPT-4. The Icelandic government recognized the critical importance of integrating its language into rapidly advancing AI technologies, and the Language Technology Programme was the vehicle that made such integration a reality.

This initiative was built upon a foundation of high-quality Icelandic training data, cultivated by Almannarómur through the Language Technology Programme. Being able to offer such a comprehensive dataset was an important hook to build tech company interest in collaboration, and the quality of the dataset enhanced the capabilities of Icelandic Large Language Models (LLMs).

Big Tech companies often appear as black boxes to outsiders, with opaque organizational structures, flat hierarchies, and hard to decipher power relationships. Almannarómur’s experience navigating those complex structures and processes provides valuable takeaways. One of these was to use high-level validators to open doors and secure initial meetings. Almannarómur had the support of Iceland’s president and ministers who played a key role in initial outreach. This included the president leading an Icelandic delegation to visit Big Tech companies, and ministers sending formal letters and actively participating in tech industry events. Building on this visit, a series of follow-up meetings led to a breakthrough moment, with OpenAI becoming the programme’s first formal major big tech partnership. This was a key validation of Almannarómur’s effort to partner with tech companies, building on more than a decade of informal partnerships with companies such as Google and Microsoft. The partnership with OpenAI then served as a credibility boost that helped stimulate engagement with other companies, facilitating further collaborations. 

The success of this initiative has drawn international attention, with many small language communities seeking Almannarómur’s guidance on replicating Iceland’s approach. This led to the publication of “The Icelandic Approach: Preserving and Revitalizing Linguistic and Cultural Diversity in AI,” a report presented at an OpenAI conference in New York, further solidifying Iceland’s position as a leader in language preservation through AI.

UNESCO Coalition for Linguistic Diversity in AI

Building on its commitment to linguistic diversity, Almannarómur established a significant partnership with UNESCO to spearhead a coalition dedicated to promoting linguistic diversity in AI development. The central goal of this collaboration is to bring together stakeholders that represent low-resourced languages to jointly advocate for their inclusion in AI. The coalition will also develop a practical roadmap and toolkit to empower small nations in their engagement with major AI companies effectively. Thus, Almannarómur has translated its success in bringing the Icelandic language into AI models to becoming a champion of other less-represented languages.

The scope and potential impact of this coalition are considerable:

UNESCO advocates for language diversity and supports the preservation of languages, including indigenous languages and already has multiple efforts on this front. One existing example is the World Atlas of Languages where the organization brings together 8,324 languages, classifying their language situation (endangered, not in use, safe, etc), location where they are spoken, and type of language (spoken, sign, written). With this background knowledge, UNESCO is currently in the process of identifying the initial group of participating nations for the Coalition for Linguistic Diversity in AI.

Including a language in the coalition involves conducting readiness assessments to understand the specific AI and language technology challenges faced by these nations. A significant obstacle is the need for a big amount of quality data to train and test the technology, especially written data, which can be scarce from some endangered languages. Furthermore, the intended application of this data – be it machine translation, speech recognition, or natural language processing – will also present its own distinct set of complexities.

Beyond data and technical constraints, this coalition faces other challenges. Like many UN agencies, funds, and programmes, UNESCO may encounter difficulties due to the new US administration’s order to re-evaluate American contributions. The United States is currently revising its membership and funding to the organization. It is unclear what programmes might be affected, but this presents a challenge for the backers of the Coalition to secure operational funding.

Challenges in Partnering with Big Tech

Despite the promising partnership with UNESCO and initial successes like the collaboration with OpenAI, Almannarómur has also encountered roadblocks in its efforts to partner with major AI and technology companies.

Small organizations with limited resources, staff, and market reach often struggle to maintain consistent engagement with tech companies and, more importantly, to transform dialogue into collaborative projects.

Large tech companies frequently have complicated internal structures that are opaque from an external viewpoint. This lack of transparency results in significant barriers to effective engagement, often making it difficult to identify the appropriate contact person or team. It also makes identification of key decision makers difficult, with C-suite level engagement not always proving the most effective. Navigating these complex systems can lead to numerous redirections, consuming valuable time and impeding progress for small companies. Furthermore, the high turnover of tech workers can leave external organizations without a reliable point of contact, disrupting continuity and forcing them to restart the engagement process after losing a key ally.

Similarly, gaining a thorough understanding of specific tech products and their underlying technical architecture can be challenging for external organizations with limited resources. Internal tech company teams can facilitate this knowledge transfer in order to help build more effective collaborations that are mutually-beneficial.

Even when initial discussions with tech companies are positive, these often fail to translate into the sustained, working-level follow-up required for concrete collaborations. Nonetheless, organizations like Almannarómur, despite the difficulties, have been able to build effective partnerships through a variety of tactics.

The Big Tech Partnership Playbook

Through its determined efforts, Almannarómur has gained valuable insights into the complexities of forging partnerships with major technology players. A key realization is that a uniform or “one-size-fits-all” approach is ineffective. Each company operates with its own distinct structure, style, priorities, and internal processes, necessitating a tailored engagement strategy.

For small organisations seeking similar collaborations, Almannarómur’s successes underscore the importance of identifying and cultivating internal advocates within tech companies. In the case of Almannarómur, finding points of contact whose personal or professional interests align with the mission of linguistic diversity has been essential. Their work also demonstrates a set of best practices:

(1) Identify “partnership champions” – company staff that understand the value of building the partnership and will advocate for it internally at the company.

(2) Establish multiple points of contact within a company to mitigate the risk of bottlenecks and ensure broader awareness. This avoids there being a single point of failure for the partnership.

(3) Implement an overall engagement strategy that includes persistent follow-up mechanisms to maintain dialogue and build momentum beyond initial meetings.

(4) Align proposed partnership with tech company priorities/mission/policy objectives.

(5) Identify mutual wins, for example where the external partner can become a validator of the tech company’s approach, policy or products in some way as a result of the partnership. This holds value for tech companies that are constantly under the microscope of regulators and are in search of allies.

(6) Embrace iterative learning – continuously adapt engagement strategies based on the outcomes and feedback from each interaction.

(7) Look for ways to leverage partnerships into follow-on projects. This can be with the same partner, or it might unlock a different process altogether, like Almannarómur’s work with OpenAI led to a partnership with UNESCO.

Ultimately, Almannarómur’s work highlights the critical importance of linguistic diversity in the evolving landscape of AI and emphasizes the indispensable role that small organisations, companies, and nations must play in shaping an increasingly technological world. Moving forward, it is imperative that smaller players and technology companies discover sustainable and mutually beneficial pathways for collaboration. These collaborations can help ensure that the ongoing advancements in AI — and technology more broadly — are inclusive of diverse and rich cultural heritages.