At Google I/O, the company announced several new AI-powered tools and features, such as a new and improved version of its large language model, PaLM 2. As others in the media have noted, Google appears to be gearing up for competition from OpenAI, which launched ChatGPT late last year and has been in the spotlight ever since.
“Seven years into our journey as an AI-first company, we are at an exciting inflection point,” the company’s CEO, Sundar Pichai said in his keynote address at the event. “We have an opportunity to make AI even more helpful for people, for businesses, for communities, for everyone.”
In his keynote address, Pichai traced the company’s incremental steps toward developing generative AI applications — it started, he says, with Gmail’s “smart reply” function in 2017. That soon evolved into “smart compose,” wherein the app dynamically suggests phrases and words to use in an email.
And now, Gmail will have a “Help me write” feature, allowing users to generate unique text directly on the platform. While browser extensions like ChatGPT Writer have allowed users to employ generative AI within Gmail for months now, Google’s “Help me write” is the company’s first in-house feature to perform this function.
But it’s not just Gmail that’s getting a boost from generative AI functions — the company announced a suite of several other generative AI-powered features at the event. PaLM 2, will power the company’s own chatbot, Bard, which performs a similar function to ChatGPT, albeit with the ability to scan the internet as well.
The company also shared that the PaLM family of LLMs includes specialized models, like Med-PaLM, which is trained on medical data, for more niche functions than general-purpose models. Additionally, the company shared its PaLM API, which allows developers to build applications and other tools that utilize the company’s LLM.
“LLMs are capturing the imaginations of people around the world,” the company writes. “At Google I/O, we’re excited to share a new suite of tools that make it easy for developers to build on top of our best models.”