Menu Menu

Google launches new AI tool PaLM 2

In an attempt to keep up with the competition, Google has announced PaLM 2, a ‘next-generation language model’ that will be available for many of its upcoming products. It will be in direct competition with services like Midjourney and DALL-E.

Google is attempting to throw itself back to the forefront of AI tools and services.

Despite years of AI research and in-house work, the company has been unable to create popular or meaningful products that resonate with consumers. Two of its largest competitors, DALL-E and Midjourney, have enjoyed significantly bigger market share since their launch a few years ago.

At its annual I/O conference, Google announced PaLM 2, a ‘next-generation language model’ that will be built into 25 new products and features. It says that this new AI will ‘outperform’ other leading systems on similar tasks.

PaLM 2 will be incorporated into different services across Google’s product line-up. For example, it will be implemented into chatbots resembling other services like Snapchat’s AI model and ChatGPT. It will be able to translate languages, write computer code, and respond to images and prompts.

Slav Petrov, co-leader of the PaLM 2 project, spoke to the Guardian about the work behind the project.

‘The neural network revolution that we are now experiencing started about 10 years ago. We’re really excited to make these models available broadly externally, because we want to see what people can do with them.’

For most consumers, the first iteration of PaLM 2 they’ll be able to interact with will be Bard, Google’s internal chatbot that will soon be available across the world. Bard is currently usable in English, Japanese, and Korean. Google aims to eventually offer 40 languages.

You’ll be able to send photos across to Bard as prompts for more information, a first for Google. During its presentation, the company showed an image of a kitchen shelf as an example of AI prompting. The user asked for a recipe based on the ingredients shown in the image, which Bard was able to provide.

PaLM 2 will also be integrated into existing services like Gmail, Docs, Slides, and Sheets, using a new ‘Duet AI’ feature.

The AI will have a multitude of uses within Google’s ‘Workspace’ apps, including image creation, automatic text generation, and idea curation. It’s likely that PaLM 2 will be valuable in saving time and cutting laborious corners within the creative process.

Google’s rush to get ahead of the curve in AI development does come with potential risks, of course. During its presentation, the company noted that its new tools were intended as ‘creative aids and not intended to be factual.’

Like any new tech product, the service will be susceptible to errors, mistakes, and ill intended consequences.

This is true of all the AI products we’ve seen burst onto the scene in recent years. Incorporating artificial learning into our work flows is still very new, and we’ve yet to regulate or standardise the process in a way that protects jobs and creators. Google’s PaLM 2 is no different.

According to the Guardian, Google’s presentation was very last minute. Updates were sent to reporters hours before the main presentation during the I/O conference, suggesting that the company is prioritising industry leadership over a fully fleshed out, competent final product.

We’ll have to see if Google can really make a dent in the already busy AI market.