Jeenius CEO Kirsten Brecht Baker highlights a significant data void in AI training, noting that current models lack the authentic conversational data needed to perform effectively in diverse, real-world settings.
Apple's AirPods are named only as an illustration in a Jeenius opinion column noting that the earbuds offer live interpreting in nine languages while only 20 languages met Jeenius' accuracy threshold. No Apple figures, guidance, product launch or regulatory action appear, so there is no material financial read-through for the stock.
Jeenius CEO Kirsten Brecht Baker argues that current AI models suffer from a critical data deficiency, limiting their real-world utility in complex environments like healthcare. While models excel at text-based tasks, they lack the authentic audio and video data required to navigate nuanced, multilingual human interactions. The company is now focusing on creating specialized datasets to bridge this gap.
The company attributes the shortfall to a lack of authentic conversational data, as most models are trained on text-based internet data rather than natural, real-world speech.
It is a strategy where qualified linguists create and benchmark datasets to ensure AI models can handle cultural nuances and complex, real-world communication.