As language models continue to scale, Large Language Models (LLMs) have exhibited emerging capabilities in In-Context Learning (ICL), enabling them to solve language tasks by prefixing a few in-context demonstrations (ICDs) as context. Inspired by these advancements, researchers have extended these techniques to develop Large Multimodal Models (LMMs) with ICL capabilities. However, applying ICL usually faces two major challenges: 1) using more ICDs will largely increase the inference time and 2) the performance is sensitive to the selection of ICDs. These challenges are further exacerbated in LMMs due to the integration of multiple data types and the combinational complexity of multimodal ICDs. Recently, to address these challenges, some NLP studies introduce non-learnable In-Context Vectors (ICVs) which extract useful task information from ICDs into a single vector and then insert it into the LLM to help solve the corresponding task. However, although useful in simple NLP tasks, these non-learnable methods fail to handle complex multimodal tasks like Visual Question Answering (VQA).

Up until six months ago, artificial intelligence might not have squeaked into the top five areas of climatetech for most people. But the Cambrian explosion of large language models -- led by ChatGPT -- has suddenly hooked hundreds of millions of users, offered mind-boggling creative capabilities that have surprised almost everyone, and kicked off an AI arms race in the tech world. What are the most compelling applications for AI in energy? This week, we feature a live conversation with Priya Donti, MIT professor and executive director of Climate Change AI; Amy Francetic, managing general partner at Buoyant Ventures; and Jesse Morris, CEO of the Energy Web Foundation. In this episode, recorded at Greentown Labs in the Boston area, we explore the wide range of AI applications for grid modeling, renewable energy integration, research & development, and product development.

When looking for a date online, it can be difficult to get a proper idea of what potential suitors are really like just by looking at their photos. But that could be set to change, with the introduction of a new video feature on Bumble. The dating app is set to launch BumbleVID, allowing users to create a story of 10-second videos, which will delete after 24 hours. Users will record 10-second videos directly within Bumble, which are posted to their profile. The videos can be viewed by anyone who comes across the profile while swiping.

It may sound like the premise of a science fiction film. But, believe it or not, scientists at Boston University claim to have discovered a game-changing blood test that could help predict lifespans. The study, published in the journal Aging Cell on Friday, used biomarker data collected from 5,000 blood samples and analysed it against the donors' health developments over the subsequent eight years. Together, they identified patterns which indicated both good and bad futures. In all, the researchers generated 26 different predictive biomarker signatures.

Artificial Intelligence (AI) is back in the headlines as more and more companies like Amazon, Google, Facebook, Twitter and Microsoft have started to commercialize its use. Four important developments in the past few days on this front caught our eye. The most recent was last week's announcement of an AI partnership by Google, Facebook, Amazon, IBM and Microsoft. The new alliance is dedicated to advancing public understanding of the sector, as well as coming up with standards for future researchers to abide by. The other important development was the release of a research report by Forrester – The Top Emerging Technologies to Watch: 2017 to 2021.