Today in AI: Google to overhaul Assistant with AI, Instagram tests content labels for AI and more

Google to overhaul Assistant with generative AI, layoffs likely

Some work on the revamp has already begun and Google is going to reorganize the teams that work on and maintain Assistant. It has already begun doing away with some roles.

• According to a report in Axios, which reviewed an internal email sent to employees, some work on the revamp has already begun and Google will reorganize the teams that work on and maintain Assistant, which will likely lead to layoffs.


• The mail sent out by Google VP Peeyush Ranjan and Director of Product Duke Dukellis said there is "profound potential" in generative AI "to transform people's lives and that they see a huge opportunity to explore what a supercharged Assistant, powered by the latest LLM technology, would look like. (Some members of the team have already started working on this, beginning with mobile)".

Instagram working on labels for 'content generated by Meta AI'

While we don't know what Meta has planned in terms of consumer AI technology, we can make a fair assumption that the launch of generative AI tools may not be far off.

• According to screenshots shared by developer and reverse engineer Alessandro Paluzzi on X, the label will inform users that they are viewing content generated by "Meta AI", though it appears that the creator will have to flag them.


• The social media giant recently unveiled the Large Language Model (LLM) Llama 2, in partnership with Microsoft. Though the model is listed as open-source, Meta will charge companies with 700 million or more monthly users a fee for a license.

Google's new AI model allows robots to finish tasks on their own

Google's Deepmind AI division calls it a vision-language-action (VLA) model that learns by scouring the web and parsing through heaps of robotic data.

• In a blog post by the team, they said RT-2 builds on the work done for RT-1 which was trained on "multi-task demonstrations", and learned through "combinations of tasks and objects seen in the robotic data".


• RT-2 improves upon these capabilities allowing it to interpret new commands and respond to user instructions using "chain-of-thought reasoning".


• The team says RT-2 can perform "multi-stage semantic reasoning like deciding which object could be used as an improvised hammer (a rock), or which type of drink is best for a tired person (an energy drink)".