Bard and ChatGPT: the problems with both
Google has made a big move in the race to create generative artificial intelligence (AI) by announcing that it is working on a programme called “Bard” that will compete with ChatGPT. Google said that its chatbot would be released “in the coming weeks.” This was a response to ChatGPT, a popular AI chatbot created by OpenAI, a modest San Francisco firm, and Microsoft.
After Google’s revelation, the race is on to discover what many believe will be the future of Internet search. It follows rumours that the largest web search company is anxious that ChatGPT, which is growing in popularity, may beat it to the punch.
But the company hasn’t said anything about plans to combine Bard with Google Search results.
So how is Google’s Bard going to work?
When people type in questions, the service will answer them in text using artificial intelligence, just like ChatGPT does. Google wrote that Bard can help plan a baby shower, review two Oscar-nominated films, and teach a 9-year-old about NASA.
Sundar Pichai, the CEO of Google, said that Bard, which is already available to “trusted testers,” is meant to put “the breadth of the world’s knowledge” behind a conversational interface. Bard is built on Google’s 2021 LaMDA generative language model for discussion applications. This means that Google Assistant will be able to talk about any topic.
“This much smaller model needs a lot less computing power, so we can add more users and get more feedback.” Google said, “We’ll use outside feedback and our testing to make sure Bard’s answers meet a high standard for quality, safety, and being based on real-world information.” Pichai also said that Google plans to give developers access to the underlying technology through an API, just like OpenAI does with ChatGPT, but he did not give a time frame.
What is the biggest difference between Google’s Bard and ChatGPT?
It looks like Google has a secret weapon to beat ChatGPT: the ability to pull information from the Internet. In a blog post, Google said that Bard uses information from the web to give fresh, high-quality answers. ChatGPT has impressed people with its ability to answer complicated questions, even if the answers aren’t always right. Its biggest weakness may be that it can’t get real-time information from the Internet.
ChatGPT’s language model was trained on a massive dataset to generate text depending on input, but the data only goes to 2021. Google demonstrated that Bard will consider multiple viewpoints when there is no obvious answer.
For example, if someone asked, “Is it easier to learn the piano or the guitar?” the answer might be, “Some people say the piano is easier to learn because the finger and hand movements are more natural… Others say that learning chords on the guitar are easier.
With the race to make AI-based chatbots that can answer questions heating up, what are the worries?
Experts say Google and OpenAI’s text production software, while interesting and well-written, is often inaccurate. Real-time Internet searches, including hate speech and racial and gender stereotypes, may detract from these new products.
Google has been more cautious with Bard than OpenAI, possibly because the big firm has more to lose. Google AI researchers cautioned against text generating technology in a 2020 draught document. This angered corporate officials, who fired Timnit Gebru and Margaret Mitchell, two prominent researchers.