SlatorPod #249 How to Expand in AI Data Services with DATAmundi CEO Véronique Özkaya
May 6, 2025
Véronique Özkaya, Co-CEO of DATAmundi and former leader at Summa Linguae, shares insights on her company's evolution toward AI data services. She highlights the increasing demand for specialized data for AI, emphasizing the complexities and challenges in the industry. Özkaya discusses the importance of blending linguistic expertise with tech capabilities to tackle issues like data scarcity and bias. Looking ahead, DATAmundi plans strategic acquisitions to bolster consultative services, positioning itself as a key player in the growing AI data landscape.
AI Snips
Chapters
Transcript
Episode notes
Core AI Challenges Explained
- AI challenges include data scarcity, hallucination, and bias, which require specialized solutions beyond simple volume labeling.
- Language service providers with deep tech and linguistic expertise can uniquely address fine-tuning needs in multilingual and domain-specific AI projects.
Human-in-the-Loop to Fix Bias
- For bias correction in machine translation, they use humans in the loop to create gender-neutral outputs, such as multiple profession sentences.
- This iterative human-involved approach reduces model bias and enhances data quality for AI systems.
Data-for-AI Market Growth
- The data-for-AI market grows at 25-30% annually, much faster than traditional language services at around 5%.
- Demand is rising for fine-tuning pre-trained models, not just large-scale crowd-sourced data collection.