Google strikes back at Microsoft with AI "experience" for search results

Tech-giant arms race heats up as Google announces AI enhancements to Search at flagship developer event
Deena Theresa
Google unveiled its most advanced language model at Google I/O 2023.
Google unveiled its most advanced language model at Google I/O 2023.


In a bid to regain ground on AI-rivals Microsoft, Google is adding a generative AI "experiment" to its search services.

According to an announcement made at Google i/o, people who opt in to Search Generative Experience (SGE) will soon see "AI snapshots" appearing alongside results from Google Search.

It seems no-one at Google's flagship developer event is listening to Geoffrey Hinton, who recently expressed concerns about the pace of AI-development, because the focus of CEO Sundar Pichai's keynote was very much AI-driven.

Used in search, the company says, AI technology will improve the accuracy of results and provide a more seamless and efficient user experience. Google's goal is to make search more intuitive, interactive and engaging for users.

Last time Google AI chatbot Bard made a public appearance, an onscreen blunder wiped about $US100 billion from parent company Alphabet's market cap.

The company will be hoping this latest move puts it back on the front foot. Early jumps to Alphabet's share price following the announcement indicate that the markets have responded well to the move.

AI development race

The release of ChatGPT by Microsoft-controlled OpenAI kicked things off back in November 2022. When Google countered ChatGPT with the release of Bard, Microsoft announced it was going to try and bring search also-ran Bing back to life by arming it with OpenAI's ChatGPT.

With a renewed focus on AI, combined with Google's dominance in the search space, it seems unlikely that this experiment with generative AI will be going away anytime soon.

As a result, it looks like the way in which we find content and information online is about to undergo a radical shift.

PaLM 2: Multilinguality, Reasoning and Coding

Also announced at Google i/o was PaLM 2, a "state-of-the-art" learning model with improved multilingual, reasoning and coding capabilities.

According to a blog post, the upgraded version of PaLM has advanced capabilities to understand, generate and translate nuanced text. The result of being trained heavily on multilingual text, covering more than 100 languages.

Google says PaLM 2 is well ahead in logic and common sense reasoning. It can interpret, generate and debug code for over 20 programming languages, ranging from Python and Javascript to Prolog, Fortran, and Verilog.

Bard is opening up to the general public for the first time and rolling out globally. PaLM 2 can be experienced through Bard.

The Mountain View-based Tech giant has released PaLM 2 in four sizes from smallest to largest: Gecko, Otter, Bison, and Unicorn. Gecko is small enough to run on a smartphone. The lightweight model is "Fast enough for great interactive applications on-device, even when offline. This versatility means PaLM 2 can be fine-tuned to support entire classes of products in more ways, to help more people," the company said.

Google has also begun work on its next model to enable future innovations. Called Gemini, the company claims that it has begun to exhibit multimodal capabilities unseen in previous models. Like PaLM 2, Gemini will be available in various sizes and capabilities to be used across different products, applications, and devices.

This article has been updated since it went live.

More on the announcements at Google i/o from IE tomorrow.

Add Interesting Engineering to your Google News feed.
Add Interesting Engineering to your Google News feed.
message circleSHOW COMMENT (1)chevron
Job Board