Google Begins UK and US Roll Out of ChatGPT Competitor, Bard

Thom Carter

21 March 2023, 04.30pm

Google Begins to Roll Out Bard in the UK and US Google sandbox

In a bid to outdo Microsoft in the big tech generative AI race, Google has begun to roll out limited access to its ChatGPT competitor, Bard.

At the time of writing, Bard is available to select UK and US users, with all members of the public in both countries now able to join a waitlist. While the roll out is currently limited, Google has stated that access to the generative AI chatbot will be available to more countries and languages over time.

Today’s roll out is big news in the global tech sphere, especially since Bard, shortly after being unveiled last month, mistakenly presented inaccurate information in a demo question about the James Webb Space Telescope — causing Google parent company Alphabet to lose around $100 billion in value.

What is Bard?
Bard, at its core, is a chatbot — but a more intelligent, AI-powered one. With it being powered by Google’s LaMDA (short for “Language Model for Dialogue Applications”) technology, the chatbot attempts to provide users with high-quality, text-based answers to their queries and requests.

The use cases for which Bard can be used have huge variance. For instance, it could be asked questions like “which Scots tech companies secured the most funding in 2022?,” while it could also be used, as Google suggested, “to give you tips to reach your goal of reading more books this year” or “explain quantum physics in simple terms.”

However, much like other recent generative AI chatbots — not least OpenAI’s GPT-4 — the answers presently supplied aren’t always factually accurate. Speaking of which, earlier today, Axios reported that GPT-4 readily spouts misinformation.

No wonder, then, Google has stated the following on the blog post for Bard’s UK and US roll out: “While LLMs are an exciting technology, they’re not without their faults. For instance, because they learn from a wide range of information that reflects real-world biases and stereotypes, those sometimes show up in their outputs. And they can provide inaccurate, misleading or false information while presenting it confidently. For example, when asked to share a couple suggestions for easy indoor plants, Bard convincingly presented ideas… but it got some things wrong, like the scientific name for the ZZ plant.”

How Does Bard (Currently) Work, Exactly?
According to those who witnessed a Bard demo, the chatbot’s user interface is much like its competitors’: the user is presented with a text box, and Bard invites them to ask questions. It then generates a few different responses for the user to choose from. Underneath the responses, there’s a “Google It” button, which navigates the user to Google’s search engine.

In the blog post, Google has also stated that they think of Bard as a “complimentary experience” to Google Search, and has been designed in such a way with the “Google It” button. This, theoretically, should help users to verify the legitimacy of Bard’s answers, before potentially believing them to be wholly true or acting on them.

The accuracy of Bard’s responses is, apparently, hit-or-miss — with the chatbot supplying three differently incorrect answers on a tricky, numbers-heavy question during the demo. When AI does this — confidently gives a response, thinking that it’s right but is ultimately wrong — it’s referred to as a “hallucination.”

Considering these hallucinations, Google has specified in its roll out post that Bard is an “early experiment,” and that the language model tech that underpins generative AI currently contains flaws.

What’s Being Done When It Comes to Ethics and Safety?
Google says that its work on Bard has been guided by its AI Principles, which are: be socially beneficial, avoid creating or reinforcing unfair bias, be built and tested for safety, be accountable to people, incorporate privacy design principles, uphold high standards of scientific excellence, and be made available for uses that accord with these principles.

The company also says that “we continue to focus on quality and safety.” As well as using human feedback and evaluation to guide their work — which this cautious UK and US roll out should surely help to support — Google has “built in guardrails, like capping the number of exchanges in a dialogue, to try to keep interactions helpful and on topic.”

As the company said themselves: “Bard is an experiment and may give inaccurate or inappropriate responses. You can help make Bard better by leaving feedback.”

To sign up for the Bard waitlist, click here.