Blog

Google, at the I/O 2023, emphasized artificial intelligence more than anything, and also launched PaLM 2—its next generation language model with improved multilingual, reasoning and coding capabilities—which also now powers Google Bard.

Google claims that PaLM 2 is “heavily trained on multilingual text,” and can function in more than 100 languages, and this ability empowers its ability to understand, generate and translate nuanced text.

PaLM 2 also has strong reasoning abilities with a wide-ranging dataset. Additionally, PaLM 2 was pre-trained on publicly available source code datasets, making it great at popular programming languages such as Python and JavaScript, and other like Prolog, Fortran, and Verilog.

“We’ll be making PaLM 2 available in four sizes from smallest to largest: Gecko, Otter, Bison and Unicorn. Gecko is so lightweight that it can work on mobile devices and is fast enough for great interactive applications on-device, even when offline,” Google said. 

The company aims to use the versatility of the PaLM 2 model and help it integrate in a plethora of whole product classes, and make it more helpful for people.

Google has successfully incorporated the advanced PaLM 2 LLM into 25 of its core products, including Google Bard and Google Workspace apps. In addition, Google’s health research teams have developed Med-PaLM 2, which is trained with medical knowledge to provide insightful summaries and answers to complex medical queries based on a range of dense medical texts.

Acknowledging the impact of AI models, Google said, “versatile AI models reap real benefits for everyone. Yet just as we’re committed to releasing the most helpful and responsible AI tools today, we’re also working to create the best foundation models yet for Google.”

Google is currently developing Gemini, the next generation AI model, which they plan to eventually replace PaLM 2 with in certain products such as Google Bard, once it is deemed ready for widespread use.