Ali Farhadi and AI2 Profiled in the New York Times
At the core of Ali Farhadi's vision for AI2's future is openness. “We’re pushing for a glass box. Open up the whole thing, and then we can talk about the behavior and explain partly what’s happening inside.” In a conversation with the New York Times, Ali talks about his career journey, the OLMo project, and more.
"It's becoming clearer and clearer that datasets are one of the most important components in training language models (and other similar models, e.g. language-to-vision). However, this data is huge, and we lack the technical tools to process it and the language to discuss it." AI2 Young Investigator Yanai Elazar partnered with collaborators from AI2 and multiple universities to attempt to identify what data is included in the enormous datasets used to train LLMs. Access the demo below, and read the paper here.
In association with Wildlife Protection Solutions (WPS), and the Wildlife Conservation Society (WCS), EarthRanger has launched Gundi, a first-of-its-kind universal adaptor for conservation technologies. The free and open-source platform aims to ensure frontline conservationists have the tools they need to rapidly scale their efforts to address growing threats to nature.
Are you a researcher hoping to improve your speed reading? Semantic Scholar's new skimming feature in its Semantic Reader aims to help by pre-highlighting and color-coding research papers.
Congratulations to Patrick Beukema, lead software engineer for AI2's Skylight team, on his inclusion in Business Insider's AI 100 2023! Read about his work and view the rest of the list at the link below.
