Google has offered some of its US- and UK-based Pixel superfans early access to Bard, the business’s retort to ChatGPT.
The select users can try it out and offer feedback to train Bard in real-life use.
Those without an exclusive invite can sign up for a waitlist for when Bard reaches the next stage of its rollout.
What Is Bard?
Like ChatGPT, Bard is a generative AI engine powered by LLMs, including Google LaMDA.
Many may remember news swirling around LaMDA last year when a Google employee raised concerns that it may be sentient.
Nevertheless, Google continues to test and finetune the model, which this next stage of the rollout will help with.
As it does so, Google is tagging Bard as a predictive engine, generating replies to user prompts.
In a way, the search engine leader already offers similar capabilities. Consider how it suggests ways to finish a sentence in Gmail and Google Docs.
However, Bard promises to be much more advanced. Indeed, it answers user questions, helps users to generate ideas, and makes lists.
After receiving a response, users can ask follow-ups to delve deeper and engage in a fully-fledged conversation, as Bard maintains conversational context.
Google Is Positioning Bard Differently from ChatGPT
Google promotes Bard as an assistive tool to "boost your productivity, accelerate your ideas, and fuel your curiosity," highlighting how the generative AI may supplement its search engine.
Current users can also follow a path to the existing search engine to look for additional answers.
Yet, in pushing the ideation and curiosity motif, Google indicates that Bard aims to support creativity, generate snippets of information, and serve up suggestions.
Unlike ChatGPT, it is not a tool to write entire essays – which is fast becoming a significant concern in the educational space.
Moreover, it doesn’t always generate an answer when it doesn’t know. In these scenarios, it will often state: “I can’t help you with that.”
However, that does not mean it will not churn out incorrect information every now and then (more on this in the section below).
In addition, Bard does trail ChatGPT-4 in several ways. For instance, GPT-4 can mimic the user's style, analyze images, and generate code.
Now, Google has announced plans to enable Bard to do the latter, make it available in additional languages, and enable voice-based applications. Yet, ChatGPT already enables the latter, as the following video exemplifies.
Nonetheless, Google has underlined its commitment to keeping all its work with Bard grounded in AI principles. Indeed, it has built guardrails into Bard that prevent it from sharing illegal, harmful, and sexually explicit information.
Additional emphasis on this narrative comes after Microsoft endured bad press after laying off its ethical AI team soon before releasing GPT-4.