What's new in AI this week? Google brings Gemini Pro to tons of new tools, and more

Welcome to What’s New in AI, our weekly update where we bring you all the latest AI news, tools, and tips to help you excel in this new brave new AI-driven world.

AI in the news: Google dominates once again

Last week Google was in the news for the release of its new LLM, Gemini. A week later Google is the focus once more, as they announced several new products and tools that utilize its Gemini. While most of Google’s news dropped on December 13, a few days before that the company announced NotebookLM is now available for general use. The tool uses Gemini Pro and is designed to give users critical insights about the notes they take.

While that’s the most mainstream-friendly of its announcements, there were quite a few exciting pieces of news for developers this week. First, Google Duet AI for Developers is now generally available. This tool makes it easy to generate code and offload certain development tasks to AI. Even more exciting, developers will now be able to integrate Gemini Pro into their apps. Last week Google released Gemini Nano to developers for on-device use through AICore. Now devs will be able to do the same with Gemini through the new Google AI Studio.

We’ll talk more about these tools in our Tools & Apps Spotlight, but that’s far from everything Google announced around AI.

As previously promised, Google has also now brought Gemini Pro over to its Vertex AI. On a related note, Google also introduced Imagen 2 for text and logo generation. The tool remains exclusive to Google Cloud customers using Vertex AI who have been specifically approved for access.

Last but not least, Google is also hoping to make an impact on the medical industry with its new MedLM, a generative AI model optimized for medical use. Two different models exist, each with different use cases. One model is larger and can better handle complex tasks, while the other is a smaller model that is for scaling across specific tasks. As Google puts it in its blog post:

“Through piloting our tools with different organizations, we’ve learned that the most effective model for a given task varies depending on the use case. For example, summarizing conversations might be best handled by one model, and searching through medications might be better handled by another.”

The exact tasks the model will perform will depend largely on the medical practice or niche involved. As Google mentioned, one MedLM user has been using it to help draft patient notes at emergency department sites. In other words, it’s mostly about reducing the labor involved with routine paperwork and other tasks that can be outsourced to AI relatively easily.

Whew! Google sure hogged the news cycle this week but let’s also run you through a quick recap of some of the other biggest AI news of the week:

• Is ChatGPT getting lazier? Maybe! More reports are coming in suggesting the model is refusing to do work it once did with ease, instead prompting users to do it themselves.
• Is manipulating a photo with AI a potential ethics issue? That’s what C. Scott Brown asked when he first played around with Magic Editor. He asked Google if they had any plans to add watermarks or any way to tell if a photo has been tweaked by AI tools. Its answer was less than clear.
• Google Gemini is powerful, but is it as good as it seems? Google released a hands-on demo during Gemini’s launch, but as it turns out the video was modified to make it look more responsive than it would be in reality.
• Is DIA ready to embrace AI? The Defense Intelligence Agency (DIA) is reportedly finalizing its official strategy to “control the use of technology for decision support and human-machine teaming when providing intelligence to warfighters and policymakers.” Read more at the Federal News Network.
• Van Gogh back from the dead? The Musée D’Orsay in Paris now has a lifelike AI version of the Dutch artist that can interact with guests and offer insights into his own life and death.
• Another new language model? While not quite as monumental as the release of Google’s Gemini, Microsoft is also releasing a new version of its small language model, Phi-2. It looks to be a major upgrade.

AI Tools & Apps Spotlight

In our AI Tools & Apps spotlight we shine the light on new apps and tools that we think are worth the extra attention. This week we’re shining the spotlight on three new tools from Google:

Google NotebookLM

Google NotebookLM is an experimental note-taking app for journalists and other writers, and it is now available for general users in the US. The new AI-powered app uses Gemini Pro AI to better understand your notes and make connections that will help you when researching a story.  For example, it might suggest follow-up questions for an interview, based on your recent conversation history and the content of your sources.

It also will give you suggested actions and advice such as telling you to combine several notes into a unified single note, critiquing your notes to help you with your prose or argument, the ability to build an outline out of the notes, and much more.

I used NoteLM to plan out a feature I was working on earlier this week and found the suggestions to be pretty helpful. I’m still playing around with it to see if it’s something I’ll use regularly, but there’s potential here.

Not only are the AI features nice but the actual UI and app are equally great, making it easy to organize your sources and other notes. I recommend giving this a try, whether you’re working on an essay or research paper, or are an online journalist.

You can check it out for yourself by visiting Google’s official NotebookLM website.

Google Duet AI for Developers

The Duet AI assistant is Google’s take on the Github Copilot, another tool that uses AI technology for code completion and generation.  The tool is designed to make suggestions and to help speed up the coding process.

The assistant is available across multiple integrated different development environments and the IDEs can integrate with Cloud Shell Editor, PyCharm, Visual Studio, and a few others. Duet also supports more than 20 programming languages including Python, Java, C, and C++.

Right now you can get Duet AI for Developers for free but Google plans to start charging in February of 2024. The service will cost $19 per user per month with an annual commitment. Head over to Google’s site to try it out for yourself now.

Google Generative AI Studio

Google AI Studio has formally arrived, providing a tool that easily lets you develop prompts and Gemini-based chatbots. Technically isn’t a new tool, as it previously went by the name MakerSuite. This new version does offer quite a few improvements, with Gemini support being the most obvious.

Right now the tool only has access to Gemini Pro and Pro Vision, but the suite will also support Gemini Ultra in the future.

Google says this is the fastest way to build with Gemini, letting you quickly put together your projects and then generate API keys so you can integrate the features directly into the app. For those who prefer, it’s also possible to access the code and work on it in a more powerful IDE.

For now, there’s only a free tier, though a paid plan will roll out sometime next year. The current free version has a generous quota as you can place 60 requests per second. You can try it right now as part of Vertex AI.

How-to & tips: Video Spotlight

Looking to learn more about AI, how to make better use of AI tools, or how to protect your privacy from AI? Each week we’ll bring you a new how-to guide or tip we feel is worth sharing. This week Gary brings us a great video on how to use ChatGPT and GPT-4 in Python. Check it out: