Last Week on AI - no. 23

We are currently observing a significant shift in the way models are evolving into multimodal products and services. The initial LLMs that we were so excited about in early 2023 are now beginning to integrate images, sound, music, and web browsing capabilities. This will serve as the groundwork for the development of highly capable agents that can assist people in their daily lives. So, let's dive in the latest news:

Open-source LLMs are getting really good. Last week Mistral AI 7 announced their 7B instruct and we are seeing lot of hype around it. It outperforms others of its size and is released under the Apache 2.0 license, with no restrictions on use.

#ChatGPT can now browse the internet to provide you with current and authoritative information, complete with direct links to sources. It is no longer limited to data before September 2021.

Tesla AI bot "Optimus" can now sort objects autonomously. Its neural network is trained fully end-to-end: video in, controls out. #optimusbot

Amazon Bedrock is now generally available for the public. Also, Meta's LLama 2 13B and 70B parameter models will soon be available on Amazon Bedrock. Securely integrate and deploy generative AI capabilities into your applications using the AWS services you are already familiar with. #aws #bedrock

OpenAI is beginning to roll out new voice and image capabilities in ChatGPT. They can offer a new, more intuitive type of interface by allowing voice conversation or show ChatGPT what you’re talking about.

Sam Altman, (CEO at OpenAI) trolled Reddit saying AGI was achieved internally. Of course we when AGI is achieved it will not be announced with a reddit comment…

John Ive (ex Chief Design Officer at Apple) and OpenAI wants to create a new AI-powered phone. The project aims to create a more natural way to interact with AI, fueled by over $1 billion in funding from Softbank's CEO Masayoshi Son.

Last week on AI is a weekly recap of the most significant?#ai?news from past week, curated by the team at Leniolabs_