It felt like a typical week. But it brimmed with numerous announcements about the Generative AI field that deserve some acknowledgment, and now it has filled this second post. We were getting breathless recounting the torrent of announcements related to Generative AI on Thursday morning Asia time (https://bit.ly/3DcoNtN). There was an avalanche on Friday.
https://www.techopedia.com/how-ai-can-revolutionize-the-drug-discovery-process
Google Bard can now see through Google Lens. It can talk, and that too in 43 languages. The multimodal features can create a diverse array of outputs, from recipes to codes or apps, from images. Meta's CM3leon is impressive under the hood (https://bit.ly/3DzaCiN). Superficially, it is another text-to-image generative model with an added image-to-text feature. But its use of transformer language models, rather than diffusion methods, sets records for efficiency and low computational cost (https://bit.ly/3NXbgeJ). By some accounts, its image generation surpasses Dall-E 2 and Midjourney, while its ability to describe images in natural language could potentially revolutionize image search.
More is going on beyond the headline-grabbing chat-product developments. https://bit.ly/3NVQqfp is a new case study demonstrating ChatGPT's utility for improving medical record traceability and retrieval. Elsewhere, Mayo started testing Med-PaLM in its research hospital, showcasing generative AI's promise for boosting healthcare productivity.
NVidia continues investing more in biotech companies working on AI-enabled drug discovery. https://bit.ly/3XQIQaF is an understandable article about https://bit.ly/3pJGnlU groundbreaking recent study and the rapid advances in the drug discovery process. It is not just about synthetic molecules: at MIT, https://bit.ly/3XV0DNN, new methods are investigated on synthetic proteins.
A survey reveals that synthetic data is utilized by 90% of the automotive industry, making it clear why AWS has established a new data center specifically for generative AI applications in the sector (https://bit.ly/44QjSKZ).
There are new investigations and regulations. New language learning models (LLMs) continue to emerge, with JD.Com being the most recent entrant. However, the landscape isn't without its threats. Cybersecurity firms are identifying malicious AI entities such as WormGPT and PoisonGPT.
There is no desire to summarise the events here on a weekly basis. The purpose of the above is to underscore the sweeping, cross-sectoral, and relentless innovation wave currently unfolding in the realm of generative AI. #ai #innovation #bard #CM3leon #healthcare #pharma
