By now, we’ve all had a conversation with a chatbot, a computer program that operates through AI (artificial intelligence) and NPL (natural language processing) to communicate with customers and potential customers. The program can understand a customer’s questions that are typed into a pop-up box and then provide automated responses.
Chances are you’ve experienced this through OpenAI ChatGPT. Not to be left behind, Google has released its own version, “Bard” to a limited group for testing. Google had to take a huge leap with Bard, and it fell pretty hard with an AI hallucination. In “human talk”, that means Bard told lies.
That doesn’t mean we should all forget Bard though! Nor should we cancel ChatGPT accounts either.
If you’d like to try Google Bard for yourself, access is only available to a few select insiders. This means you’ll need to get on the waitlist by going to Bard’s homepage and click on the “Join waitlist” button. It will advise you if your account is eligible and then you’ll be asked to receive Bard news updates.
What’s next? Wait. Wait until Google advises you that it is now your turn to use Bard. The Bard homepage has a notice at the top stating this is “Bard Experiment” and the FAQ page admits when the new tab opens that “some of the responses may be inaccurate”.
There is a clear warning stating that Google’s own views are not necessarily represented by the answers and information provided by Bard. But that isn’t just Google’s Bard; these same warnings are now universal with AI chatbots on other pages using AI chatbots.
Bard: What is it like to use this AI chatbot?
About the same as with other chatbots, but maybe not as skilled yet. For example, the word choices are repetitive, yet it is highly creative and generates ideas. However, that creativity included blurting out details that had the potential to be messy using speculative prompts.
As you have a conversation with Bard, you could choose three different possible automatic responses. This feature reminds us of Dalle-2 by Open AI. It generated content in the same manner so that every output was less definitive and not a final answer. Resulting in responses that are hallucinated and come with two more attempts to get the right response.
Google’s Bard, occasionally, will quote responses with excerpts, much like the AI chatbot on Microsoft’s Bing. And when it comes to difficult questions, it has noticeable struggles, as other chatbots do.
Looking to the future, some say that chatbots could make a company money. But officials at Google say that Bard is too young to plan on that for now. In the meantime, Google generated over $160 billion last year from ads that were placed next to Google search results according to The Wall Street Journal. This would imply that before you and Bard can become friends, you’ll be interrupted by ads that can suggest ways to save on your home insurance.