Google Bard vs Microsoft Bing Chat: Which AI Is Smarter?

Those keeping a keen eye on AI these days will know of the constant launches by different companies. From Microsoft launching the new Bing AI Chat, and OpenAI coming out with GPT-4 to Google Bard AI, it’s mayhem. However, among so many options, it can be difficult to separate the wheat from the chaff. While these AI chatbots come with their own capabilities, some definitely do the job better than others. We earlier compared Google Bard vs ChatGPT and came out with interesting results. However, we are here today comparing Google’s first-ever chatbot against Microsoft Bing AI Chat, a GPT-4 powered search engine that has become massively popular. So, which AI should you rely on for your day-to-day needs, and which one has the better features? Let’s not waste any more time and dive in to find out.

Google Bard vs Microsoft Bing: AI Search Engine Battle (2023)

We have compared both AIs on various parameters and listed their major features and differences. Depending on which feature you’d like to see first, use the table below to jump to the necessary section.

Google Bard vs Microsoft Bing: Key Differences

As mentioned above, both chatbots come with their own certain advantages. Let’s discuss which AI search engine has the better features and their major differences.

1. Based on Separate Models

Both Google Bard and Microsoft Bing are AI chatbots that have been trained on datasets. To simplify it, the bots have essentially scrapped this data and gained information from it. They then use the data to come up with answers to different questions and scenarios users create. However, there is a fundamental difference between Bard’s and Bing’s language models.

Courtesy: Google

For starters, Google Bard is based on a smaller-sized version of Google’s own language model called Language Model for Dialogue Applications, or LaMDA for short. Google’s language model has been fine-tuned for conversations, creative outputs, and coming up with text that doesn’t exist anywhere else. As Bard itself states, the training data for LaMDA is a mixture of text and code, including books, articles, websites, and more. Google claims that the model can “can engage in a free-flowing way about a seemingly endless number of topics,” which gives it more conversational power than Bing.

On the other hand, Bing is based on the company’s Prometheus model, which we know is Generative Pre-Trained Transformer 4, or GPT-4 for short. GPT-4 is OpenAI’s latest and most advanced LLM yet, with significant improvements over the last iteration. GPT-4 is a model that has been specifically trained to have a sharp reasoning response. However, the company has also paid attention to the nuance of conversations, and as such, the model has been trained on a fine set of human conversation data too. Further, GPT-4 encompasses more advanced features such as being multimodal and having language support.

While there is no clear-cut winner in this scenario, GPT-4 does possess some distinguishing features that make it stand out.

Winner: Undecided

2. Google Bard Plagiarizes Text

Even though Google’s LaMDA model can supposedly come up with brand-new text and engage in conversations, it’s having trouble. In this particular instance, we are referring to our last conversation with Bard in the comparison linked above. When asked about Luffy and Haki in the One Piece anime, Google Bard plagiarized a few lines from Beebom’s article word-for-word. It did not even provide any sources, which is very surprising.

Unfortunately, we are not the only ones this has happened to. In a recent tweet, Tom’s Hardware’s Editor-in-Chief Avram Piltch showed how Bard had plagiarized an article from their website for its own output. While Bard did correct itself after being prompted, plagiarizing content like this indicates a flaw in its system.

Microsoft Bing, on the other hand, did not copy any texts when asked about different topics. We also ran various texts through our list of the best AI plagiarism checkers and no exact match was found. Furthermore, if you ask Bing Chat to write you an essay in creative mode, it refuses and says that would be plagiarism and unethical.

This is an amazing addition to Bing Chat and a welcome one at that. In our experience, the same sadly cannot be said for Google Bard. As this has happened multiple times, we hope Google acknowledges this misstep and puts proper guardrails in place to ensure publications and independent creators do not suffer. It would be great if AI chatbots didn’t copy off texts, actually rephrased the texts, and linked the source in the footnotes (more on this below).

Winner: Bing

3. MS Bing Cites Sources

It’s good to see Microsoft Bing maintain originality even after being asked a variety of prompts. However, another thing Bing does correctly is that it provides linked citations to its inspirations. This gives the user the advantage of always going back and checking if the bot has copied anything. MS Bing continues to provide sources no matter what you ask it and even embeds them into text.

This is a stark contrast to Google Bard, which gives you no sources whatsoever in most cases. Speaking on its lack of citations, Google Bard AI mentions that it doesn’t do that because it’s not an academic researcher (huhh?). However, Google Bard also mentions that it can only provide information it has been trained on and cannot guarantee the accuracy of the same. While this does not stop the chatbot from hallucinating, it still is a good sign that Google has taken proper steps with Bard to warn users about its potential mistakes.

While you should never fully rely on an AI chatbot like ChatGPT for your full research, MS Bing’s inclusion of sources is still a step toward making AI more reliable. So if you ask us, Bing wins this one.

Winner: Bing

4. Google Bard Is Much Faster

Google has always been known for its lightning-fast response time. No matter what you search, the search engine dishes out responses in a second. Google Bard AI, fortunately, follows in its footsteps. From the time you search for a query, Bard takes a little over 3 seconds to come up with a full-fledged response. This includes complex outputs such as essays, poems, jokes, and even more. Bard AI’s extremely fast response time was very impressive when it came to quickly wanting an answer.

Contrastingly, Microsoft Bing AI is slower. The bot goes through the general ropes, and you can almost feel it formulating an answer. The responses too are generally slowly typed out instead of being given all at once, like in the case of Bard. This gives Bard a considerable speed advantage over Microsoft’s search engine. While Bing does its due diligence, it clearly doesn’t do it fast enough.

Speaking from a purely impatient standpoint, if speed is the primary thing you are looking for from your AI bot, Google Bard is clearly the one to prefer.

Winner: Google Bard

5. MS Bing Can Generate Images

Google Bard can do a lot of things with its LaMDA model but creating images isn’t one of them. However, Microsoft’s AI Bing has recently received the ability to take in prompts and generate images. Powered by OpenAI’s DALL-E model, the Bing Image Creator can natively take in image prompts and come up with beautiful and polished images. This makes the AI search engine the one place to generate both visual and textual content.

Note: this feature isn’t live for all stable users just yet

However, it’s worth noting that the Bing Image Creator hasn’t rolled out everywhere just yet. However, even for times like those, Bing can still search the web for images and describe them to you. Furthermore, you can even ask for image prompts to use in some of these best AI art generators. If asked, Bing can even go into more detail and give detailed prompts to use in the generators.

Google Bard has given no indication as to its capability of generating images. There might be support added in the future, but for now, the Bing Chat feature of Microsoft’s search engine can do both in this Bard vs Bing battle.

Winner: Bing

6. MS Bing Can Code

When Google launched Bard, one of the FAQs mentioned that the bot cannot help with coding just yet. While support might be added in the future, it’s not here now. Nonetheless, we put Bard through some coding rounds, and it produces basic enough code to copy off. Sadly, ChatGPT found some of this code broken which fits what Google mentions. Furthermore, even though Bard claims it can code in various languages, putting it in a complex situation messes it up completely.

Microsoft Bing, on the other hand, is much better at coding scenarios imagined by users. This extends to website and app creation, specific requests, and more. Bing was also able to accommodate our request to ‘Write a shell script to compress PNG files in a folder with zopfli,’ which is a handy prompt we found on Twitter. While Bing won’t fully be able to help programmers with complex situations, it will do a far better job of it when compared to Google Bard. For those who prefer their search engine to have coding abilities, Bing passes with flying colors.

Winner: Bing

7. Bing Has Longer Memory

One of the best things about ChatGPT is that it can remember a lot about previous conversations. While the bot does not exactly use past responses to come up with new ones, it remembers roles. This means you can assign it the role of a spell checker, a coder, and much more, and it will adopt its persona immediately.

Microsoft Bing follows suit with Bing Chat’s long memory. When assigned the role of a spell checker, Bing AI happily agreed to check the spelling of whatever you fed it. The bot remembered its role for multiple turns and kept correcting my spelling throughout my use of it. This extends to different personalities so you don’t have to be restricted to just spell check. In this way, MS Bing can be fine-tuned to suit a user’s exact needs and become sort of a personalized chatbot.

Google Bard, however, has horrible memory right now. When asked to correct my spelling errors, the chatbot did agree. However, Bard immediately forgot its role when given a paragraph with incorrect spellings. It did not resume its role and had to be reminded about the same. Fortunately, Google has already mentioned this and said that “Bard’s ability to hold context is purposefully limited for now.” However, as the bot continues to learn, this should change and maybe we will learn more about its capabilities at Google I/O 2023.

While Google Bard gives a good glimpse into its helpful nature, it is not to be relied upon when giving it personality or remembering past conversations.

Winner: Bing

8. Bing Has Different Modes

Speaking of fine-tuning, Microsoft didn’t stop at giving its chatbot just a long memory. The Bing Chat feature also comes with different conversational styles that change the bot’s tone and verbosity. Bing comes with three different styles, titled “More Creative, More Balanced, and More Precise.”

True to their names, Creative makes the bot imaginative. Balanced provides a fine balance between creativity and facts. Finally, Precise makes Bing AI all business and starts a concise and straightforward chat. Depending on what your use case, you might be inclined to pick one use case over another. Microsoft’s extension of conversational variety is a big boost to getting the right answers.

Google Bard sadly is limited when it comes to different conversational styles. However, that is not to say the bot isn’t friendly or helpful. It does its best to answer your questions and actually comes up with good and helpful outputs. However, the chatbot seems to be limited to a single mode for now, and given the fact that its memory isn’t exactly pristine, this limits the things you can do with it. As spotted by Twitter user Peter, here are the default settings for Google Bard:

Image Courtesy: Twitter/Peter

Combined with its ability to remember assigned roles with its different conversational styles, Microsoft Bing Chat wins this particular fight.

Winner: Bing

9. Google Bard Gives Multiple Drafts

While Microsoft Bing might be more malleable, Bard also has a few tricks up its sleeve. One of the best things about Google’s new chatbot is that it comes with a feature that the company calls “Drafts.” Whenever you ask the bot any query, it outputs three different versions of the same answer. The users can read through all the drafts and decide which answer fits their query the most. While the drafts mostly have the same level of detail, there are times when a particular answer might be better than the other.

Drafts by Google Bard have been done to prevent the AI from sounding too authoritative. Since no answer is the clear-cut winner, users can mix and match these depending on what they want.

In contrast, Bing only gives one rendition of its answer, which might sound a bit absolute to people. While you can ask for another answer, it is not offered by default. Compared to Bard, this isn’t exactly the best move as people not savvy with the bot will miss out on getting more nuanced and detailed information. We are hoping that in light of Bard, Bing comes out with its own version of drafts. However, as it stands now, Google Bard’s multiple answers are a big advantage the bot holds over Bing.

Winner: Google Bard

10. Bing Is Multilingual

Google Bard isn’t exactly a search engine at the moment. However, that won’t stop people from trying to use it like one. And for a search engine to function effectively, it must support a variety of languages that include everyone. Google Bard by default only supports US English for now. The same has been confirmed by the company that mentions support for other languages will come soon. However, Bard interestingly still believes it supports other languages. But when asked to compose a poem in German, the bot messes up and doesn’t know how to.

Bing, however, has support for multiple languages. This allows Microsoft’s chatbot to compose poems, write essays, compose recipes, and more in a variety of languages. Even though GPT-4 has documented support for over 26 languages, Microsoft has not mentioned exactly which languages Bing supports. However, when asked Bing AI mentions that it can understand languages such as English, 中文, 日本語, Español, Français, and Deutsch. We also made Bing write poems in languages not mentioned and it did that just fine.

When it comes to global support, Google Bard is severely limited for now. For those looking for a search engine that is multilingual, Bing Chat is the one to go to.

Winner: MS Bing

Google Bard AI: Pros and Cons

Pros Cons
Extremely fast outputs Occasionally copies text
Gives multiple versions of the same answer Limited contextual memory
Informs user about its fallacyCannot code well
Only limited to English
Doesn’t cite sources

Microsoft Bing: Pros and Cons

Pros Cons
Cites sources for its answer Slightly slower response time
Offers different conversational styles Occasional glitches
Long contextual memory Limited in some ways when compared to ChatGPT
Can also generate images
Can perform a good level of coding

Google Bard vs MS Bing Chat: Verdict

It’s always difficult crowning a singular champion. Both Google Bard and Microsoft Bing Chat are AI chatbots capable of leaps and bounds. While Bard can give responses at lightning-fast speeds and give drafts, Bing is meticulous with sources and personalities. However, there is a fine line to cross to become the winner, and Bing does that.

Microsoft Bing has a lot of things going for it. Powered by the latest GPT-4 LLM, the AI-powered search engine-cum-chatbot is an extremely effective conversationalist. The bot combines that with its efficient data research that lets users know where the answers come from. However, Bing is more than capable of cooking up content too with the bot being able to code complex scenarios without breaking a sweat. Adding to this is Bing’s multilingual support and AI image generation which means AI goodness for everyone.

Google Bard sadly pales in comparison when pitted against Bing AI. It copies text, forgets what it’s supposed to be doing, also hallucinates, and doesn’t cite sources to boot. While we would have excused everything, an AI copying text feels like death’s kiss.

While it is understandable that Google Bard is in experimental stages and to be taken with a grain of salt, I was still expecting the bot to have a modicum of competition against giants like Bing and ChatGPT. For now, it feels like a rushed preview that could have been much better if given more time. While I am sure Bard will get better once exposed to more training and user interaction, it definitely loses the battle right now making Bing Chat the winner.

However, like most things in life, your mileage may vary. Depending on what you may want out of AI, you may find Bard to be the best AI ever created or absolute trash. What you decide on is ultimately up to you.

Google Bard vs Microsoft Bing Chat: To the Victor Go the Data

Microsoft Satya Nadella recently said in an interview with The Verge, “Look, at the end of the day, they (Google) are the 800-pound gorilla on this (search engine category). And I hope that with our innovation, they will come out and show us that they can dance.” Moreover, he added, “I want people to know that we (Microsoft) made them dance. And I think that will be a great day.” So yeah, we know at least one man who’s happy with these results.

That said, we hope you liked our in-depth comparison between Google Bard and Microsoft Bing. No matter which side you may be rooting for, I’m sure we can agree that the artificial intelligence field is crazy right now. There are so many things you can do with AI, that it’s crazy. There are all these best AI writing apps, AI music generators, cool things you can do with ChatGPT and so much more. The list is practically endless. However, with the scale of AI increasing day by day, it’s only bound to get longer. So which bot did you end up siding with? Tell us your choice in the comments below.

Comments 1
  • Ben Allen says:

    Microsoft Bing *does* plagiarize, so I’m not sure how many tests were done or how recently. I recently asked it a question about the benefits of a certain service: it used a single source, and copied all of the bullet points and much of the body text in the answer. Then I asked it a different form of the same question. Although it included 2 sources this time and did not plagiarize any body text, it still copied the bullet point text (about a sentence long each) verbatim.

Leave a Reply