Google Bard v/s OpenAI ChatGPT: Which chatbot is better
The generative AI battle between Google and Microsft has entered a new phase, with Google now opening limited access to Bard. With ChatGPT flexing its GPT-4-powered muscles, it is imperative for Google that Bard makes an impression. Although both are chatbots based on language models, there have stark differences. And that brings us to the question, which chatbot is better?
Why does this story matter?
Bard did not have the best of starts. The chatbot had a failed demo in February that cost the company $100 billion in valuation. On the other hand, OpenAI's ChatGPT has been crushing the chatbot game. However, it is not without faults, giving its rivals some hope. Google aims to capitalize on ChatGPT's imperfections with a slow and measured approach.
ChatGPT's LLM has more parameters than Bard's
Bard and ChatGPT are based on large language models (LLMs). The former is powered by LaMDA (Language Model for Dialogue Applications), while the latter is either backed by GPT-3.5 or GPT-4 (depending on the versions). LaMDA has 137 billion parameters (values the LLM can update as it learns), while GPT-3 has 175 billion parameters. They are based on Google's open-sourced transformer deep learning model.
ChatGPT's knowledge base has a cut-off date
One of the major drawbacks of ChatGPT is its inability to answer questions about events after September 2021. Bard, on the other hand, has no such restrictions. This is related to the LLMs behind both. LaMDA is capable of extracting answers from the internet, which keeps Bard up to date. ChatGPT draws its answers from GPT-3 or GPT-4's knowledge base.
Bard cannot code yet
One of the things that made ChatGPT immensely popular is its ability to write code. The chatbot has shocked both laymen and professionals with its coding abilities. It can also debug code. Bard is still learning to code. It can't help users with codes yet. According to the FAQ, "responses about code aren't officially supported now."
Bard cannot remember past conversations
ChatGPT's ability to remember past conversations is something that made it stand out from others. The feature isn't perfect, but it is much better than what other chatbots can do. Bard, on the other hand, cannot do that. Per Google, Bard's ability to hold context is "purposefully limited" for now. The company said that ability would grow over time.
Bard gives multiple responses to a prompt
One feature of Bard that could be a trendsetter for AI chatbots is 'drafts.' The chatbot creates multiple responses to a prompt and allows users to pick the best response. The feature puts the onus on the users to pick an answer rather than the chatbot picking an answer. ChatGPT, on the other hand, gives one answer at a time.
ChatGPT is better than Bard at the moment
It's time to answer the question now: which chatbot is better? It is evident that ChatGPT is ahead of Google Bard at the moment. It had a headstart, and it's making the best of it. Google is now playing catch-up. Its measured approach might be a drawback in this scenario. In a few months, we will know how good Bard can get.
