Like Microsoft, Google can’t stop its cloud from pouring AI all over your heads

Google popped a bunch of AI models onto its cloud platform on Wednesday for folks to try out and perhaps adopt.

One being Gemini Pro, the text-generating system teased last week to developers and enterprises. The other models are Google’s picture-generating Imagen 2, and a family of medical-related generative AI tools dubbed MedLM.

These products follow the launch of Gemini, a collection of content-generating models powered by what’s claimed to be Google’s most powerful large language model architecture yet. This also comes after Microsoft and other big names in cloud and business IT, including Google, rushed to flood their wares with what’s marketed as machine-learning enhancements – a theme for 2023 and likely 2024 and sadly 2025.

Gemini comes in various sizes, from Nano for on-device workloads to Ultra for heavy lifting on backend servers. The mid-sized Gemini Pro, unveiled last Wednesday, can now be accessed in preview form via an API provided by Google Cloud’s Vertex AI: developers can use this remote interface to build their own homegrown chatbot applications on top of Gemini Pro, or so we’re told. Programmers can adapt the model by carefully engineering its prompts, fine-tune it on their own data, and connect it to other APIs to shape its behaviors and capabilities for specific tasks.

If you want to bake a natural-language interface into your app, you could use Gemini Pro, just like you could use OpenAI’s ChatGPT, etc.

Google also today said Duet AI for Developers, a chatbot service hoped to boost programmers’ productivity (your mileage may vary), is now generally available. It’s the usual programming assistant affair. We’re told it works with various IDEs, and attempts to do things like autocomplete source code as developers type, answer coding queries, help troubleshoot, and offer guidance on how to use third-party software from MongoDB, Crowdstrike, and others.

In fact we’re told more than 25 suppliers have worked with Google to ensure their products are understood and covered by Duet AI for Developers.

“For example, a developer writing code using MongoDB will be able to ask Duet AI for Developers, ‘Filter customer orders over $50 in the past 30 days by geography, and then calculate total revenue by location,’ and Duet AI for Developers will then use information from MongoDB’s products to suggest code to complete the task, so developers can build even faster,” Google veep Gabe Monroy explained

Over the next few weeks, the large language models powering all of the Duet AI services will be upgraded to Gemini, too. The Developers service is currently free to use up until January 12, 2024.

Duet AI in Security Operations is also now generally available; this is a chatbot service built for handling queries about protecting infrastructure, analyzing network logs, and so on.

Enterprises looking to use AI for visual tasks, such as creating digital art or logos, can potentially turn to Imagen 2, now generally available via Vertex AI. The text-to-image tool was developed by engineers at Google DeepMind, and the latest version is better at generating photorealistic pictures and rendering text more accurately to promote brands, according to the bumf. It can also write captions and answer questions about images.

Social app Snapchat, graphic design platform Canva, and stock image site Shutterstock are already using Imagen, we’re told. All images produced by Imagen 2 will contain a SynthID digital watermark. These are said to be invisible to the naked eye and can be detected computationally to identify images as synthetic.


You get a Copilot, and you get a Copilot – Microsoft now the Copilot company


Finally, Google has launched MedLM, a class of large language models focused on the medical uses for healthcare. There are two models, both are based on the Big G’s Med-PaLM 2 system.

One is larger and more powerful than the other, and is designed for more complex tasks such as sifting through academic papers and documents to generate leads for potential new drugs. The other can handle easier chores, such as summarizing conversations between doctors and patients, and medical question and answering.

Early adopters of MedLM models include clinic HCA Healthcare and drug designer BenchSci, as well as Accenture and Deloitte. ®


This website uses cookies. By continuing to use this site, you accept our use of cookies.