Google unveiled a slew of AI features at its latest I/O conference, with new, improved AI models to be integrated across all of its major products.

Tech giant Google announced a plethora of new Artificial Intelligence (AI)-backed features during its annual Google I/O conference, with updated AI tech set to appear across its major platforms.

On May 10, the annual Google I/O conference took place in California, with CEO Sundar Pichai giving a keynote address on the most significant updates to the firm’s AI stack, among other announcements.

Google shows its hand

The Pathways Language Model (PaLM) was revealed by Google last August. Since then, developers have used the language learning model to ship generative AI-related apps such as ever-popular chatbots.

Google updated its model with “PaLM 2,” boasting improved capabilities around reasoning, coding and multilingualism as the model was trained on more complex and varied subject matters.

PaLM 2 will come in various sizes — with one iteration of the model able to deploy on mobile phones.

Google said the new PaLM is the backbone for over 25 apps and showcased two specialized models. Med-PaLM 2 for medical applications and Sec PaLM 2 for use in cybersecurity.

Bing gets bested? — Google Search soon to get AI backup

Google’s flagship Search product is getting AI backup as rival Microsoft’s Bing beat Google to the punch when it integrated OpenAI’s ChatGPT as part of its search engine.

The feature, matter-of-factly called the Search Generative Experience (SGE), will have a limited experimental launch to opted-in U.S. users for testing before Google considers a wider rollout.

The tool seemingly collates information from webpages and delivers it in a ChatGPT-like response in Google Search above actual search results according to demonstrations of the product.

SEO is about to change forever.

Above the fold in Google search results will be completely filled by generative AI result. pic.twitter.com/K5xDII0bBi

— Pete (@nonmayorpete) May 10, 2023

It also provides information about searched products when users use the “shopping” option in Search. In the Google-provided demonstration, the model gave advice on what aspects to look for in a bicycle when the user searched for e-bikes for example.

Google’s Bard gets souped up

Bard — Google’s answer to ChatGPT — was part of the products that received the PaLM 2 treatment, getting souped-up features and a broader rollout to boot.

The conversational AI model was launched around two months ago but only in the U.S. and the United Kingdom, now it’s been rolled out to over 180 countries with more to come.

Part of Bard’s upgrade includes improvements to its coding abilities and repertoire, Google also improved its citations with the bot highlighting where it fetched certain code from.

Generative image AI tools from graphic software company Adobe will also soon be integrated into Bard, allowing it to generate images from a prompt akin to similar popular tools.

Mail, Docs, Maps and more gets AI-ified

Many other Google products also got the backing of PaLM 2 and Pichai ran through a series of demonstrations showcasing new AI-powered features on Google’s Maps, Docs, Mail and Photos.

One demonstration highlighted an AI-charged version of Gmail’s “Smart Compose” feature that can automatically generate email responses using a prompt. It can also be refined to make the text more formal, elaborate, or brief and seems to pull data from the email thread to bulk out responses.

A similar product, “Magic Compose,” is coming to Google’s Android phones with AI-generated responses that apparently help make a message transmit the “desired vibe” such as “chill” or “Shakespeare.”

Soon, @Android will be able to give your *compose-itions* an extra spark of personality. Magic Compose, a new Messages feature powered by generative AI, helps you transform your text message to your desired vibe before you hit send → https://t.co/ZkpSB2E7Hd #GoogleIO pic.twitter.com/GCoqXbIFBB

— Google (@Google) May 10, 2023

Gemini could palm off PaLM with even newer AI

Despite just launching PaLM 2, Google is also working on an apparently more advanced large language model called “Gemini” to replace it or, at least, provide another option.

Gemini is still in training but Pichai said Google is “already seeing impressive multimodal capabilities not seen in prior models.”

He added once it’s “fine-tuned and rigorously tested for safety” then Gemini, like PaLM 2, will also launch in various sizes and capabilities.

Source: Cointelegraph.