Ecolinguistics And AI

<p>In the last few months since&nbsp;<a href="https://medium.com/@george.vallego/ecolinguistics-and-ai-70145b84f362" rel="noopener"><strong>Part II</strong></a>&nbsp;of this series, various models have emerged that, despite being smaller than GPT and Bard, exhibit similar outstanding performance. Notably, these include Llama, Claude2, Mistral, and Zephyr. We extended our ecolinguistic exploration to encompass all of these models and found that there is a significant lack of ecological alignment in their discourses as previously shown for GPT4 in&nbsp;<a href="https://medium.com/@george.vallego/ecolinguistics-and-ai-d1d46efbe72b" rel="noopener"><strong>Part I</strong></a>. Furthermore, we discovered that in some cases, this gap is even more pronounced in non-English languages.</p> <p>These discoveries have led us to the H4rmony project, which is dedicated to realigning AI with ecological values and principles through the application of ecolinguistics, facilitated by&nbsp;<strong>Reinforcement Learning</strong>.</p> <p><a href="https://medium.com/@george.vallego/ecolinguistics-and-ai-6e00e4787833"><strong>Read More</strong></a></p>