Alright folks, Google is making some moves in the realm of generative AI. They’re rolling out a bunch of AI models and tools across their Workspace apps and Cloud. And get this, they’re expanding their personal virtual assistant, Duet AI. This chatty bot can do all sorts of things like generate material for Google Slides, organize data in Sheets, and even create backgrounds in Meet.
Now, Google isn’t here to mess around. They want to bring Duet’s generative functionality to other areas of Workspace. So basically, you ask it to do something, and it’ll do its best to make it happen. Pretty cool, right?
But here’s the kicker, my friends. Even though Google practically invented the neural network architecture behind generative AI, they’ve been seen as falling behind Microsoft and OpenAI. These two have been hyping up their AI products, billing themselves as pioneers, while Google has been flying a bit under the radar. It’s a classic case of not shouting about your work from the rooftops, and then playing catch-up. They’ve got some ground to make up in the public’s eye.
But fear not, my friends. Google is not one to back down from a challenge. They’ve just announced that Duet is making its way to enterprises, priced at $30 per month. No word yet on pricing for us regular folks.
If you’re an IT department looking to give Duet AI a whirl, you can request a trial. Powered by Google’s PaLM 2 language model, this tool can generate text in Google Docs or Gmail when given an input prompt or query. It’s also great for diving into Google’s security-related services, like Mandiant Threat Intelligence.
And that’s not all, my friends. Duet AI can also produce images and captions in Google Slides, organize data in Sheets, take notes in Google Meet, translate speech in multiple languages, and even chat with users as a chatbot in Google Chat. It’s a real multitasking champion.
Now, let’s talk updates. At their Cloud Next ’23 event, Google announced that PaLM 2 now supports a whopping 38 languages. They’ve also extended its context window, which means it can process even more text in its input prompt. It’s like giving a language model an extra boost of brain power.
But that’s not all, folks. Google has also made improvements to Codey, their code-generating model, and Imagen, their text-to-image software. They claim the quality is better than ever.
But wait, there’s more. The folks over at Google DeepMind have come up with a clever tool called SynthID. It watermark images generated by Imagen in a way that’s invisible to the human eye but can be detected by software. It’s all about detecting those fake AI-generated images, my friends.
Right now, SynthID is being released to a select few customers who use Imagen on Google’s Vertex AI platform. Gotta stay one step ahead of the AI fakes, you know?