ChatGPT’s new competitor What is Google Bard and how does it work?

Developed by Open AI, ChatGPT has become so popular that some have begun to claim that it will replace the Google search engine in the near future. Could the great features of ChatGPT really replace Google? This question is still unknown, but one of the technology leaders, Google, has entered the competition of chatbots that can create humanoid texts using Bard AI.

Developed by Google, Bard AI is a chatbot designed to complement the company’s search engine. Bard uses the Google LaMDA language model, which is similar to the ChatGPT GPT-3.5 language. This means that he will give answers to many topics in a nutshell and can create completely new texts in his answers that are not found anywhere else.

What is Google Bard AI, how does it work and how is it used? What exactly makes the Google Bard AI Chatbot different from ChatGPT and other alternatives you can use today? Here’s everything you need to know about Google Bard AI…

What is Google Bard AI?

Google Bard AI is a productive conversational AI chatbot similar to ChatGPT and Bing Chat. The company says that Bard can help with creative tasks, explain complex topics, and generally extract information from a variety of sources online. It can also answer tricky questions, like finding recipes that match the ingredients in your fridge. You cannot answer such questions today using just a Google search.

Bard transforms Google from a humble search engine into a capable virtual assistant. In an official blog post, Google asked Bard’s chatbot to “learn more about the best strikers in football right now and then practice their skills” or “explain the new discoveries of NASA’s James Webb Space Telescope to a 9-year-old kid.” mentioned that it can be used.

Beyond the simple questions, “Which is easier to learn, piano or guitar, and how much practice does it take to do both?” You can also ask questions such as: Answering these open-ended questions can be difficult, even for a human, and requires at least a few minutes of research. Google claims that Bard can fit information from dozens of web pages into just a few paragraphs. Based on images and demos, these answers may appear at the top of Google search results pages. However, currently accessing Bard requires visiting a specific ChatGPT-like website.

How does Google Bard work?

As mentioned earlier, the Google Bard chatbot has some similarities with ChatGPT. Both use a broad language model at their core and are optimized for open conversations. However, they do not use the same language model. Bard uses Google’s own LaMDA model, while ChatGPT uses the GPT-3.5 and GPT-4 models. Bard uses a different language model than ChatGPT, and this model can affect how well Bard works.

The problem with major language models like LaMDA and GPT is that while they can render text fairly well, the quality of the responses varies greatly. The ability of chatbots to generate text depends on the data. For example, ChatGPT information data expires in 2021. In other words, if you ask about events that happened after 2021, they may respond with completely fabricated information. Likewise, any deviation in the data can lead to skewed responses.

These two limitations probably explain why it took Google so long to deploy Bard AI against long-established competitors like the GPT models developed by OpenAI. The bard can generate answers that are potentially incorrect but appear to be correct, which can lead to the spread of misinformation. In fact, it has already happened. Google shares plunged when the company released an official promotional video in which Bard makes a factual error. So it’s no surprise that since Google Bard became available, the company has added a plethora of disclaimers.

You may be interested in:

All urban legend! 8 misconceptions about technology

What is LAMD?

LaMDA stands for Language Model for Dialogue Applications. It is a machine learning model trained on text samples, optimized for word and sentence prediction. Thus, it is a chatbot that generates human dialogues.

Google used its own neural network architecture based on an open source converter to create LaMDA and developed a model for the dialogue. Meanwhile, OpenAI uses the same converter architecture for the GPT family of language models. Google used a 1.56 trillion word dataset for LaMDA that was derived from “general data of conversations and other public web documents.” This means that the company has prepared LaMDA specifically for oral assignments.

According to a Google Research blog post, LaMDA has three main goals: quality, safety, and reliability. These goals help the chatbot generate responses that sound logical and interesting in the context of the prompt. For example, LaMDA never responds with something generic like “I understand” or “I’m glad to hear that.” In order to improve its true accuracy score, Google has also made it possible for LaMDA to search for information from external sources. In other words, he can search the web in real time to expand his answers.

How to use Google Bard AI?

After several weeks of silence, Google finally granted access to Bard. However, before interacting with the Bard, you should sign up for the waiting list and hope for the best. In addition, the Google Bard AI chatbot is currently only available in the US and UK. It is not yet clear when other regions and languages ​​other than English will be available.

Given the huge demand for chatbots, it’s no surprise that Google initially restricted access to Bard to a small number of users. After all, Microsoft did the same when Bing Chat first came out.

As with other machine learning-based chatbots, Bard AI is likely to cost Google dearly in terms of computing resources. By some estimates, each chatbot response costs ten times more than a regular call. Limiting the visibility and use of Bard to a small number of users can help the company scale those costs over time.

You may be interested in:

The 8 Best AI Rendering Apps

Bard AI and ChatGPT: which is better?

The biggest difference between Bard and ChatGPT is that Google’s chatbot has access to live information on the web. This is very different and important from ChatGPT, which only has access to the text of training data collected up to 2021.

Bard is also suitable for various non-search Google products such as Android, Chrome OS, Gmail, Google Docs, and even the Chrome web browser. Microsoft has already integrated Bing Chat into the Edge web browser, Skype chats, and the Teams app. Google has announced that it will provide AI support for Workspace and Gmail, but no details have been released.

But with these integrations, you can probably ask Bard to generalize websites and create documents directly. You can also interact with the Bard using your voice. This is a feature you cannot do with ChatGPT.

Another difference has to do with language abilities. With Bard, we will have access to the Google LaMDA model for the first time. So far, almost all AI chatbots (including ChatGPT and Bing Chat) have relied on some variation of the OpenAI GPT-3.5 and GPT-4 models instead of LaMDA. Google’s model may work just as well or much better, but we’ll only know for sure as Bard becomes more mature.

Finally, the Google Bard chatbot gives shorter answers to help users find what they are looking for faster. This means that ChatGPT is a better choice than long replies when you need to write a long article or email.

Source: 1