IS4: Human Intelligence in the Age of Multilingual AI

Presented by: L’Meese Greaney, Translated

Artificial intelligence is advancing rapidly, yet large language models reflect an incomplete worldview. Truly multilingual LLMs, built to serve global communities, depend on high-quality multilingual data. This need creates a new role: professionals who combine linguistic expertise with data annotation and evaluation skills. Their work addresses a core risk of AI adoption: fluent outputs masking errors in register, genre, pragmatics, or cultural nuance. Translation and localization bring rigorous disciplinary standards that protect human intellectual agency, ensuring AI systems support informed judgment rather than replacing it with homogenized output. This session examines how multilingual experts apply judgment to AI training data.

Takeaways:

1. Why multilingual human expertise is critical for pre-training LLMs.
2. How human judgment safeguards intellectual agency.
3. Gain insight into emerging professional opportunities and industry implications.

IS4: Judgment Over Fluency: Managing Risk in AI-Driven Localization

Presented by: Petra Angeli, Acolad

AI has turned the localization landscape into noise: more output, more speed, more “good enough” fluency — and far more risk. In this chaos, the real value isn’t perfect sentences; it’s the human judgment that decides what is safe to release and what isn’t. This MicroTalk shows why hiring for judgment, not fluency, is the only way to restore order. Petra Angeli shares practical frameworks for when to trust models, when to escalate, and how to protect brand integrity in AI-heavy workflows. The message is simple: technology generates content; humans decide whether it deserves to see the light of day.

Takeaways:

1. How judgment — not linguistic polish — becomes the gatekeeper of safety, trust, and brand protection in AI-assisted localization.
2. What “order” actually looks like: clear escalation paths, risk thresholds, and human-in-the-loop checkpoints.
3. How to identify and hire people who can manage chaos pragmatically — not get overwhelmed by it.

IS4: Indianization Through Process: A Workflow-Driven Approach to Succeeding in India

Presented by: Mayank Das, Lexiphoria

This MicroTalk introduces a workflow-driven approach to Indianization, showing how global brands can succeed in India by moving beyond translation and adopting a structured process built on research, strategy, localization, cultural adaptation and analytics. The session explains how understanding India’s diverse micro-markets enables more relevant content, stronger user engagement and measurable impact. A brief case example demonstrates how applying this process improved regional resonance and conversions for a global product launch. Attendees will gain practical insights they can immediately apply to navigate India’s complexity using a scalable, culturally intelligent localization workflow instead of isolated content adaptations.

Takeaways:

By the end of this session, attendees will learn:
a. Why traditional translation-only localization fails in India – and why India’s diverse micro-markets require a deeper, process-driven approach.
b. A clear understanding of the five Indianization workflows including research, strategy, localization, cultural adaptation, and analytics.
c. How to integrate cultural and behavioural insights into content, UX, messaging, and regional strategy for higher resonance.
d. How to embed Indianization into global workflows so teams move from ad-hoc localization to a structured, repeatable process.
e. Practical steps to achieve scalable, long-term success in India using a blend of strategy, localized content, cultural intelligence, and data feedback loops.

IS4: Learnings from the development of HaaF (Human as a Feature) components integrated with TaaF (Translation as a Feature) models

Presented by: Antonio Tejada, LanguageLine Solutions

The rise of native AI translation features in enterprise applications (TaaF) creates new challenges in embedding human-in-the-loop processes directly within those platforms. I will share lessons learned while enabling HaaF (Human as a Feature), including integration interface design, touchless content workflows, editing environment requirements, workforce management adaptations, and flexible billing models. These insights offer practical guidance for implementing human-in-the-loop AI workflows, highlighting not only the technical considerations but also the impact on language-team operations and the importance of effective Change Management.

Takeaways:

Challenges of Human in the loop models within native AI translation features in enterprise software.

IS4: Technology-Driven Language Preservation: A New Horizon for South Asia’s Indigenous Tongues

Presented by:  Shafi Ullah and Irfan Ullah, Abasis Linguistic Services

Emerging technologies offer a transformative pathway to preserve South Asia’s indigenous and endangered languages. Through digital documentation, AI-driven tools, and innovative linguistic solutions, communities can safeguard their cultural identity while making rare languages accessible to future generations. Technology bridges gaps in education, research, and communication, enabling sustainable preservation efforts across Afghanistan, Pakistan, and India. By integrating modern tools with community engagement, we can ensure these unique languages continue to thrive in the digital age, fostering diversity and strengthening regional cultural heritage.

Takeaways:

Understanding the landscape and challenges of rare and indigenous languages across Afghanistan, Pakistan, and India.
Exploring how emerging technologies—AI, NLP, and digital tools—are transforming language preservation and documentation.
Learning practical strategies and real-world examples to support, sustain, and revitalize endangered languages through innovation and collaboration.

IS4: Women, fire, and dangerous things: Can we really achieve gender-neutral language?

Presented by: Marina Pantcheva, RWS

Gender is encoded in language in many ways: from systems that group women, fire, and dangerous things to those where even inanimate objects require masculine or feminine agreement.
These grammatical distinctions are deeply embedded in linguistic structures. Yet fairness and inclusivity increasingly demand gender-neutral expression.
Is it possible to achieve this without sounding artificial or forced?
In this short talk, I outline strategies for neutralizing gendered languages and discuss what’s realistically achievable. In doing so, I examine how AI can be applied to produce gender-neutral language: where it falls short, where it succeeds, and why.

Takeaways:

The attendees will:
1. learn how gender and noun-class systems differ across languages
2. understand why it’s difficult to achieve gender-neutral language while maintaining naturalness
3. learn where large language models succeed and where they fall short in producing gender-neutral language