This paper tries to quantify the ethical dilemma of using culturally toxic training data to improve the performance of AI tools for ultra low-resource languages such as Indigenous languages. Our case study explores the use of Bible data which is both a commonly available source of training pairs for translators of Indigenous languages and a text which has a trail of physical and cultural violence for many Indigenous communities. In the context of fine-tuning a WMT19 German-to-English model into a Guarani Mbya-to-English translator, we first show, with two commonly-used Machine Translation metrics, that using only Bible data is not enough to create successful translators for everyday sentences gathered from a dictionary. Indeed, even fine-tuning with only 3,000 pairs of data from the dictionary produces significant increases in accuracy compared to Bible-only models. We then show that simultaneously fine-tuning with dictionary and Bible data achieves a substantial increase over the accuracy of a dictionary-only trained translator, and similarly happens when using two-step methods of fine-tuning. However, we also observed some, measurable, contaminated text from the Bible into the outputs of the best translator, creating concerns about its release to an Indigenous community. We end by discussing mechanisms to mitigate the negative impacts of this contamination.

In this paper we study the fine-tuning of pre-trained large high-resource language models (LLMs) into many-to-one multilingual machine translators for extremely-low-resource languages such as endangered Indigenous languages. We explore those issues using datasets created from pseudo-parallel translations to English of The Bible written in 39 Brazilian Indigenous languages using mBART50 and WMT19 as pre-trained models and multiple translation metrics. We examine bilingual and multilingual models and show that, according to machine translation metrics, same-linguistic family models tend to perform best. However, we also found that many-to-one multilingual systems have a tendency to learn a “rogue” strategy of storing output strings from the training data in the LLM structure and retrieving them instead of performing actual translations. We show that rephrasing the output of the training samples seems to solve the problem.

We investigate native language identification (LangID) for Brazilian Indigenous Languages (BILs), using the Bible as training data. Our research extends from previous work, by presenting two analyses on the generalization of Bible-based LangID in non-biblical data. First, with newly collected non-biblical datasets, we show that such a LangID can still provide quite reasonable accuracy in languages for which there are more established writing standards, such as Guarani Mbya and Kaigang, but there can be a quite drastic drop in accuracy depending on the language. Then, we applied the LangID on a large set of texts, about 13M sentences from the Portuguese Wikipedia, towards understanding the difficulty factors may come out of such task in practice. The main outcome is that the lack of handling other American indigenous languages can affect considerably the precision for BILs, suggesting the need of a joint effort with related languages from the Americas.

In this paper we explore the improvement of intent recognition in conversational systems by the use of meta-knowledge embedded in intent identifiers. Developers often include such knowledge, structure as taxonomies, in the documentation of chatbots. By using neuro-symbolic algorithms to incorporate those taxonomies into embeddings of the output space, we were able to improve accuracy in intent recognition. In datasets with intents and example utterances from 200 professional chatbots, we saw decreases in the equal error rate (EER) in more than 40% of the chatbots in comparison to the baseline of the same algorithm without the meta-knowledge. The meta-knowledge proved also to be effective in detecting out-of-scope utterances, improving the false acceptance rate (FAR) in two thirds of the chatbots, with decreases of 0.05 or more in FAR in almost 40% of the chatbots. When considering only the well-developed workspaces with a high level use of taxonomies, FAR decreased more than 0.05 in 77% of them, and more than 0.1 in 39% of the chatbots.