Google Bard

Abhishek Maheshwarappa
3 min readApr 3, 2023

Experimental conversational AI service

Hey there! I’m excited to share my initial experience with Google Bard. I finally got access after being on the waitlist, and I couldn’t resist tinkering around with it.

So, let me tell you about my impressions and thoughts on this intriguing Conversational AI chatbot called Bard.

Photo by Mojahid Mottakin on Unsplash

Google has been developing this exciting experimental conversational AI service that uses LaMDA, which stands for Language Models for Dialog Applications. When OpenAI released ChatGPT, it created quite a buzz and quickly gained a massive number of users in a short period. So, it was only natural for Google to respond with their creation, Bard, another AI chatbot designed for engaging conversations. It’s mesmerizing to see how these tech giants are competing to bring us the best AI experience!

Google Bard utilizes a Transformer neural network as its foundation. You see, Transformer models are a special kind of neural network that excel in handling tasks related to natural language processing. The way it works is quite fascinating — first, the input text is converted into a series of vectors. Then, these vectors are employed to produce another series of output vectors. Finally, the output vectors are transformed back into the desired output text. It’s a remarkable process that allows Bard to generate impressive music compositions!

Bard is quite similar to ChatGPT in that it takes a prompt as input and generates an output based on that prompt. It just takes a few seconds to respond, and often provides alternative draft answers.

However, when it comes to having human-like conversations, Bard doesn’t quite measure up to ChatGPT, which does a better job in this regard. Bard displays the final output all at once on the screen, while ChatGPT typically types out the response without delay. In my opinion, typing out the results word by word creates a more human-like experience.

As for the output content, it’s more or less comparable to Chat GPT in most general cases or questions. It’s hard to judge Bard’s quality since it’s still in the experimental stages.

One neat feature is that Bard references the sources of information it uses, adding a level of authenticity and verification to the information provided.

Obviously, it solves basic mathematical problems correctly. However, the best part is how it explains why running 100 miles in 8.3 hours is difficult, citing the need for breaks, refueling, and even discussing the training aspect.

In conclusion, Bard is a fresh addition to the cutting-edge AI technology landscape, reflecting a significant shift in the field of AI.

The remarkable progress that has been made leaves me in awe. Having grown up in rural India, where access to electricity, internet, and public transportation was a challenge, I’m truly inspired by the recent democratization of the internet in my home country. Now, witnessing the emergence of incredible technologies like Bard and GPT-4 is nothing short of amazing.

Feel free to connect with me on LinkedIn

--

--