The Gist

  • All the AI updates. Google I/O, Google’s annual developer conference, took place on May 10, with a reveal of all the tech giant’s latest new products and updates. 
  • Meet PaLM 2. Google has unveiled its largest AI language model to date.
  • New AI-enhanced gadgetry. With the release of the Pixel Fold, the Pixel 7a Android phone and the Google Pixel Tablet.

Google just unveiled its latest AI advancements, product releases and innovations — including the highly anticipated launch of "PaLM 2," Google's largest AI language model to date.

As CEO Sundar Pichai took the stage yesterday for the opening keynote at Google I/O 2023, he noted that “AI is having a very busy year” — a phrase some might call an understatement.

It’s been a transformative, ever-evolving year for AI and the annual developer conference provided a perfect venue for the tech giant to showcase the launch of new AI-powered features products and updates, aimed at enhancing user experience.

Google Launches PaLM 2: Most Powerful AI Language Model Yet

With the release of PaLM 2, Google offers its largest AI language model to date, featuring more capabilities, efficiency and speed than previous models, and its now powering more than 25 Google products and features, including Bard and new Workspace features to assist with writing tasks in Gmail and Google Docs, Slides and Sheets.

Pichai credited the joint work of the Brain Team and DeepMind for the progress made on PaLM 2 and said he considers its debut to be the latest step in Google’s “decade long journey to bring AI in responsible ways to billions of people.”

“We recently brought these two teams together into a single unit, Google DeepMind,” Pichai said. “Using the computational resources of Google, they are focused on building more capable systems safely and responsibly.”

PaLM 2 is trained on 540 billion parameters and integrated with a new technique called “chain of thought prompting” that allows users to “describe multi step problems as a series of intermediate steps.” With improved multilingual, reasoning and coding abilities, PaLM 2 underwent pre-training on an extensive collection of publicly available source code datasets, enabling it to excel in programming languages such as Python and JavaScript. Additionally, it has the capability to generate customized code in specialized programming languages like Prolog, Fortran and Verilog.

It is also trained extensively on multilingual text covering more than 100 languages, giving PaLM 2 enhanced capabilities to comprehend, produce and translate complex text — it even achieved "mastery" level in advanced language proficiency exams. And its data set is vast — encompassing scientific papers, math and expressions to enhance its abilities in calculation, reasoning and logic. PaLM 2 will be available in four sizes, including a lightweight version that can run on a mobile device.

Pichai also spoke briefly about Gemini, Google’s next generation foundation model, which is still in training.

“Gemini was created from the ground up to be multimodal, highly efficient as tool and API integrations and built to enable future innovations like memory and planning,” Pichai said. “While still early we're already seeing impressive multimodal capabilities not seen in prior models. Once fine-tuned and rigorously tested for safety. Gemini will be available at various sizes and capabilities.”

Related Article: Google I/O Leaks: Are You Ready for Some Tech-Tastic Surprises?

Google Launches Pixel Fold, Android Phone and Tablet

Among the many announcements shared during Google I/O, Google unveiled its $1,799 Pixel Fold, a thin, foldable phone with a 7.6-inch screen, powered by Google’s Tensor G2 chip, now for pre-order with shipping next month.

Further, Google announced the availability of its Pixel 7a Android phone, also powered by the Tensor G2 chip with 8GB RAM. The Pixel 7a will replace the Pixel 6a, and is available to purchase now for $499.

Finally, the Google Pixel Tablet, first revealed at last year’s Google I/O event, is also now available for pre-order at $499.

Learning Opportunities

Google's AI Updates to Maps and Translate Improve UX and Assist Global Needs

From enabling flood alerts in India to helping refugees from Ukraine communicate, Pichai said Google is not only dedicated to improving user experience, but also to advancing relief efforts, noting that Google Translate has served as a crucial communication tool for many. But while he said more people are using Google Translate than ever before, he admits there is still work to do to make it universally accessible.

“There is a long tail of languages that are underrepresented on the web today and translating them is a hard technical problem,” Pichai said. “So, with advances in machine learning, we have developed a monolingual approach, where the model learns to translate a new language without ever seeing a direct translation of it.”

With this latest approach, he announced 24 new languages were added to Google Translate.

And as part of Google’s effort to “explore the world in more intuitive ways,” Pichai announced a new experience in maps with "Immersive View" that will start rolling out in select cities this year.

With new AI-powered functionality, Immersive View will not only provide users with a chosen route — it will also provide data on air quality information, weather and traffic. And if you're looking to grab a bite, users can search for nearby restaurants and then take a peek inside with advanced neural renderings that create an immersive drone-like experience.

Worldwide, Immersive View has mapped around 1.6 billion buildings and more than 60 million kilometers of roads. However, some remote and rural areas with distinct building types and terrain remain notoriously difficult to map. To address the issue, Sundar said Google incorporated computer vision to detect buildings at scale from satellite images, and as a result, increased the number of buildings on maps in Africa by five times since July 2020 — from 16 million to nearly 300 million, with plans to double the number of buildings mapped in India and Indonesia this year.

According to Sundar, international organizations like the United Nations and the World Bank are already using it to better understand population density and to provide support and emergency assistance.

“Using advances in 3D mapping and machine learning, we're fusing billions of aerial and street level images to create a new high-fidelity representation of a place,” Sundar said. “These breakthrough technologies are coming together to power a new experience.”

Maps also offers eco-friendly routing, a product launched last year that provides users with the most fuel-efficient route.

“Eco-friendly routes have already rolled out in U.S. and Canada and people have used them to travel 86 billion miles, helping save an estimated half million metric tons of carbon emissions, the equivalent of taking 100,000 cars off the road,” Sundar said. “I'm happy to share that we are expanding this feature to more places, including Europe later this year.”

fa-regular fa-lightbulb Have a tip to share with our editorial team? Drop us a line: