At this year’s I/O conference, Google has announced of a variety of AI projects, however the one catching the most attention is their prototype, Project Tailwind. This AI notebook is punctuated with a sophisticated yet unique trait, the ability to learn from your documents. Josh Woodward, the Senior Director of Product Management for Google shared that “Like a real notebook, your notes and your sources power Tailwind”. He further elaborated that this AI model enables users to directly pick the files from Google Drive and effortlessly generate a personalized and private AI model trained on the requested information.
Project Tailwind is seen as an extremely helpful tool for students. In a visual demo, Project Tailwind went through several study notes and extracted pertinent information and text such as topics, kind of questions and a smart glossary of topics. Woodward further iterated the use of Tailwind for various task such as article writing, earning calls analysis, and legal document scrutinisers.
The computation load of training Project Tailwind is generally expensive and poses as a challenge in the production of this work. Furthermore, the generated AI samples are always at risk of data bias, skewing information and leading to false results. Google is open in acknowledging these two concerns and has already opened up the opportunity to the public, to assess Project Tailwind under the new AI Labs program. On the other hand, people are actively indulging in to the development and fine-tune language models, trained on content and files available in Google Drive.
AI is swiftly revolutionizing the modern world and it is no surprise that companies are actively competing to develop products providing a competitive edge. However, caution should also be taken when using such AI based technology as at times information accuracy cannot be certain. Google, in association with eminent writers and academicians, have been leading the research of language models and Project Tailwind gives us all a better understanding of how language models are created and adapted. These use cases serve as a useful resource and further drive AI innovation in the modern world.
