Confronting Google's BARD and Microsoft's ChatGPT: An In-Depth Comparison to Find The Leading Language Model

Confronting Google’s BARD and Microsoft’s ChatGPT: An In-Depth Comparison to Find The Leading Language Model

Quick Links

Key Takeaways

Both Google Bard and OpenAI’s ChatGPT are powerful AI chatbots with unique features and capabilities. Bard stands out with its integration with Google Search, providing access to real-time information, while ChatGPT relies on its curated knowledge base. Bard excels as an AI-powered search engine, while ChatGPT shines as a conversational AI assistant.

In the arena of AI chatbots, two major players have emerged: Google Bard and OpenAI’s ChatGPT. These chatbots are designed to engage in conversation, but each offers slightly different capabilities and features. Let’s get to know them a little better.

Google Bard Explained: How Does Google’s AI Chatbot Work?

Google Bard , launched in March 2023, is Google’s answer to OpenAI’s ChatGPT. Powered by Google’s PaLM2 , Bard can draw responses from the internet, providing up-to-the-minute information. This technology also allows Bard to provide direct links to websites when prompted, a feature not seen in its competitor, ChatGPT .

At the time of writing the free version of ChatGPT has no direct internet access, nor does the standard version of the paid ChatGPT Plus version. However, ChatGPT Plus subscribers can activate a web browsing plunging as a beta feature.

That being said, the Bing AI search engine is built on GPT-4 and is also directly connected to the internet, so it’s a more direct competitor for Bard as a web search tool in particular.

Both ChatGPT and Bard have the ability to generate programming code, find bugs, and explain what a piece of code does. However, there’s a caveat for both. First, ChatGPT is not fine-tuned for coding. It has impressive coding capabilities despite this, but that’s not what it’s meant for. If you use the paid GPT-4 version of ChatGPT, things are even more robust, but OpenAI has Codex , which is specifically designed to convert natural language into code and should (on paper) outperform ChatGPT in this area.

In the case of Bard, the ability to work with code at all is a relatively recent addition, with the announcement going out in April of 2023 . I don’t see either of these services being used for professional coding in light of the existence of specialized versions of this technology meant for coding, but it’s sure to have a major impact for anyone learning to code, or who want to create software as a side project,

ChatGPT Uncovered: What Makes OpenAI’s Chatbot Stand Out?

OpenAI’s ChatGPT was introduced in November 2022 , predating Google Bard by a few months. It is powered by the Generative Pre-training Transformer-3 (GPT-3) and GPT-4, and has shown impressive capabilities, particularly in coding, but really across multiple professional spheres such as legal and medical, to name a few.

One of the key features of ChatGPT is its ability to remember previous conversations, albeit with a word limit that’s being revised as the model advances. However, it does not use past conversations to form responses. It’s also worth noting that while Bard pulls from the internet for its responses, ChatGPT’s responses come from its static knowledge base, which has a cutoff date in September 2021, limiting its access to more recent information and research. You’ll either have to feed it new information manually, or make use of the new beta web browser plugin.

ChatGPT also has, in my opinion, peerless linguistic abilities. It can generate, manipulate, and generally run rings around any task that relies on command of language.

Google Bard vs. ChatGPT Features

Since both of these AI Chatbots are rapidly evolving even as you read this, any comparison of the two is likely to become stale faster than an NFT . That being said, as of this writing, there are a few things we can say about how these bots measure up.

Related: What Is ChatGPT Plus?

Both have broadly the same capabilities, including coding, answer math problems (with limited success), reasoning through word problems, and of course writing.

As I alluded to earlier, one major difference is that Bard does not remember previous conversations. If you look to Bard’s FAQ Google answer as to why this is has to do with them taking baby steps:

Bard’s ability to retain context is purposefully limited for now. As Bard continues to learn, its ability to retain context during longer conversations will improve.

Unlike ChatGPT, Bard can respond with images as response to a prompt. Interestingly, GPT-4 (which only paying ChatGPT users get) is actually a multimodal model. Which means it can work with visual inputs and outputs as well, but currently that’s not active in its ChatGPT implementation.

Any comparison of features here is also thrown into disarray thanks to the launch of the ChatGPT Plus plugins store . With third-party services able to give ChatGPT various superpowers, it can equip up to three plugins at once, and more are being added all the time.

The Power of Google Search in Bard: A Unique Advantage?

Bard’s integration with Google Search certainly sets it apart. The ability to pull responses directly from the internet gives it access to a vast amount of information, which can be incredibly useful. However, this doesn’t necessarily mean that Bard always provides better responses.

Since ChatGPT is drawing from its own optimized knowledge base, which has also been heavily curated and reinforced by an army of human trainers, it’s often faster and more concise when all you want are answers rather than links to websites.

In any event, the web browser beta feature currently available to paying customers bridges the gap between ChatGPT and Bard. Don’t forget that there’s the Bing AI chatbot as well, which is tuned for web search in particular.

While I don’t think being Google-connected gives Bard the massive advantage people might assume, a major potential win for Google here is integration into the rest of the Google ecosystem. With Bard and other systems derived from PaLM 2 like to be integrated into Google Docs, Gmail, Google Photos, Maps, and every other Google service that so many of us rely on every day. ChatGPT will never get that level of integration into these services, which might be Bard’s unique advantage.

Comparing Bard and Bing in the Real World

Getting a definitive sense of which chatbot is better, is basically impossible. Both are capable of a variety of tasks, and even some tasks that the people who created them have no idea they can do. Because they are also constantly learning from user feedback and additional training, the frequent pace of language model updates also makes any opinion of which bot is best at one particular thing age like milk in the sun.

Often both will give responses that are perfectly acceptable, but different. The same way you can get good but different results from asking two people the same question. What is the same between both, and this will never change now matter how much they advance, is that you can never 100% trust anything that comes out of either. Even with a web connection, there’s always the possibility of “hallucination”, misinterpretation, or other shenanigans. Which means that, while both can save you time and are incredibly useful , neither can operate without human supervision.

Related: Don’t Trust ChatGPT to Do Math

Having used both, I’d generally give the edge to ChatGPT as being better as a chatbot, and conversational AI assistant, with Bard being better as an AI-powered search engine. In fact, it makes more sense to compare Bard to Microsoft’s GPT 4-powered AI chatbot than to ChatGPT.

While I have not found the web-plugin version of ChatGPT to ever go off the rails like Bing’s version did in the early days, it’s far less reliable and predictable than either the offline version of ChatGPT, Bard, or indeed Bing AI.

So despite many writers gleefully posting things like “the dad joke test” to compare these systems, the truth is that every person’s needs are so different, and these bots are so complex, that you really have to test them yourself. The good news is that both offer a free service level open to almost anyone. Which means you can put the exact same prompts into both, specific to your needs, and then judge which one gets it right more often.

Tech Haven