ChatGPT vs. Google Gemini (Ex-Bard) — Which Bot Is Better?

ChatGPT vs. Google Gemini (Ex-Bard) — Which Bot Is Better?
03 April, 2024 • ... • 86 views
Ana Balashova
by Ana Balashova

If you’ve been keeping tabs on the AI scene, you’ve likely stumbled into the “ChatGPT vs Gemini (ex-Bard)” debate. Everywhere you turn, there’s a new bot claiming to be the best. But chasing every shiny new AI could drain your budget really quickly. So, what’s the smart move? We’re breaking it down in this guide, putting ChatGPT and Google Gemini in a face-off and even getting them to spill the beans on each other. Ready for some surprises? Let’s dive in and find out which bot earns the crown.

Overview of ChatGPT and Google Gemini (ex-Bard)

Diving into the world of artificial intelligence chatbots, ChatGPT and Google Gemini stand out as two giants most often chosen by the crowd. 

According to Similarweb, OpenAI’s ChatGPT clocked in a staggering 1.6 billion visitors in February 2024, while Gemini drew 316 million. This massive gap in traffic is a sign of how users are gravitating towards ChatGPT’s robust features and versatile applications, from crafting emails to generating creative content. 

ChatGPT has become a go-to for everything from ChatGPT for email marketing to software development, thanks to its conversational AI that can juggle text prompts and spit out helpful answers powered by GPT-3.5 on a free version. Think of it as having a personal assistant who’s also a coding guru, creative writer, and research whiz all rolled into one. Plus, for those seeking cutting-edge services, ChatGPT Plus subscribers get the latest GPT-4 model, the bonus of web browsing for real-time information, interacting with image inputs, chatting away with voice commands, and using a suite of custom GPTs — plugins that extend its capabilities even further.

On the flip side, Google Bard (renamed Gemini in February 2024) allegedly brings its A-game by tapping into the search engine giant’s vast reservoir of web pages and current events. It’s like Google Search on steroids, ready to feed you the latest scoop in more personal, conversational-style responses. While Gemini might not have the same range of creative tools as ChatGPT, it excels in pulling up-to-date info straight from the internet, making it a go-to for queries needing the freshest answers. The free version of Gemini uses the 1.0 Pro AI model to craft its answers, Gemini Advanced gives you access to Google’s most capable AI model, 1.0 Ultra. It’s claimed to be better at reasoning, following instructions, coding, and creative collaboration.

So, what’s the verdict in this ChatGPT vs. Google Gemini services showdown? Stick around as we explore the differences across the various performance areas testing both chatbots.

Comparing ChatGPT vs. Google Gemini

When it came time to evaluate ChatGPT (on a paid plan) and Google Gemini (on a free plan), I asked the chatbots themselves what parameters we should consider and what advantages each one has over the other.

ChatGPT suggested it’s tough to declare a definitive winner since the tech is changing every day. I couldn’t agree more: Today’s tech big break could be tomorrow’s old news.

Thanks to its real-time web access, Gemini flexed its digital muscles and suggested that it is better at using up-to-date information and more accurate. Gemini also highlighted its unique image search capability. 

Asking Gemini how it’s better than ChatGPT. The answer states up-to-date information, accuracy, research assistant, and image search.

Now let’s put all of those and some other things to the test.

Platform capabilities

ChatGPT was developed by OpenAI and powered by the groundbreaking GPT (Generative Pre-trained Transformer) models, with GPT-4 being the latest progeny. This AI is an all-rounder capable of creating content, debugging, and even giving advice in various professional fields. It allows the creation and usage of task-specific plugins for ChatGPT Plus subscribers. A voice feature allows users to interact with ChatGPT using voice commands, and it has the “vision” to recognize data from images.

Engineered with Google’s DNA and built on the foundation of research and advancements made by Google AI teams, Gemini initially harnessed the power of LaMDA (Language Model for Dialogue Applications), which now evolves with PaLM 2 (Pathways Language Model) at its core for better context understanding.

Its knack for pulling up-to-date information directly from the web gives it an edge in delivering timely and accurate responses. The fact that Gemini is multilingual and can analyze audio, images, and video content proves its ambition to be more than just a repository of information.

Accuracy of factual information

Google Gemini has been praising itself for its real-time access to search data, suggesting it could have the upper hand in pulling up the latest stats. I have tested and extensively used both tools for months now, and I can’t agree with this judgment. More often than you’d expect, Gemini pulled up facts from the nooks and crannies of the web — the kind of web pages you’d probably scroll past on a good day. 

For this test, I started conversations with both chatbots with a straightforward question about global email traffic. Let’s look at the sources that both contenders choose and identify which would be generally accepted as “better,” by avid internet researchers. 

Google Gemini picked an outdated, less reliable source. Venngage isn’t without its merits, but in this context, it lagged behind with a Moz Domain Authority (DA) of 77 and a Trustpilot score of 2.1 from just nine reviews. Domain authority, a number from 0 to 100, is a useful comparative metric for assessing a website’s established credibility and backlink profile. It can be a good starting point for identifying the strength of the domain, but this metric can be easily manipulated with some black hat SEO techniques. And since I still had some doubts, as I’ve never heard of Venngage and DA 77 is pretty high, I went to check the company reviews on Trustpilot. A high Trustpilot score with many positive reviews suggests that the company is real and its domain is reliable. Gemini’s pick didn’t pass this test. The bigger issue was the age of the data (it was from 2023) and a missing link to the quoted stats, not exactly what you’d call best AI tools practices.

Testing Gemini factual correctness by asking How many emails are sent per day worldwide and getting the answer 347.3 billion

ChatGPT sourced the data for its answer from Oberlo, a site with a sturdier reputation (Moz DA 78 and a Trustpilot score of 2.6 based on 430 reviews). The source has more fresh data from 2024 and the link to the study was easy to locate in the article ChatGPT was referring to.

Testing ChatGPT factual correctness asking How many emails are sent per day worldwide and getting the answer 361.6 billion

So, when we’re tallying up points on the accuracy of factual information, in my opinion, ChatGPT has the edge.

Work with images

When asked how it’s better than ChatGPT, Gemini confidently stated that it can “search the web for relevant images to complement its responses,” while “ChatGPT is limited to generating AI images (restricted to paid versions).” Gemini can also generate images, though this feature was recently met with a fair share of criticism for some pretty wild historical inaccuracies, sometimes replacing White people with images of Black and Native Americans so much so that the company had to pause Gemini’s image generation of people.

So, while I couldn’t compare the differences in testing an image generation feature asking for the portrait of the first US president, I made another simple request: “Give me a picture of a cream jar to use as a header image for my beauty-themed newsletter.”

Gemini’s response: Instead of providing an actual image, Gemini offered a link to an Amazon product page displaying cream jars. While this may seem like a relevant solution, it falls short in several ways:

  • There was no image file, only the link to a product page.
  • Copyright concerns: Using an Amazon product image could raise potential copyright issues, limiting its usability.
  • Lack of customization: The image is a generic product shot, lacking the tailored feel necessary for a newsletter header.
Testing Gemini’s work with images and getting a link instead of an actual image

On the other hand, ChatGPT delivered a nice-looking, AI-generated image of a cream jar that perfectly aligned with the requested criteria. As an original creation, it eliminated copyright concerns and provided flexibility for use in various marketing materials.

I feel like ChatGPT takes the win in this round of AI capabilities. Do you agree? 

Testing ChatGPT’s work with images and getting a generated image of a cream jar with a brief explanation of why it is good to use as a header

Handling of ambiguity

When I threw the curveball question, “Whar fox ay”  both Google Gemini and ChatGPT Plus did well, showcasing their mastery in natural language processing even with difficult prompts.

Testing how Gemini handles ambiguity with two possible interpretations of a misspelled question: as a question about a song or about literal fox sounds

While both navigated the ambiguity well, ChatGPT Plus’s proactive clarification request showed a nuanced understanding and management of human-AI interactions. As a human, I quite liked that. But we can add a point for this round to both chatbots’ scoreboards. 

Testing ChatGPT handling of ambiguity and getting a one-paragraph response with information about a song and an ask to clarify the request at the end of the response

AI detection

For this test, I asked both ChatGPT and Google Gemini to give me their best shot on a hot topic: the latest AI trends. I took the outputs and ran them through ZeroGPT, a free and simple AI detector.

Asking ChatGPT to write an intro into a post about the newest AI trends for AI detection test
Asking Gemini to write an intro into a post about the newest AI trends for AI detection test

The style turned out to be quite different. Gemini’s version had too many exclamation marks and no substance, while ChatGPT’s version was plagued with well-known AI clichés. We, those standing on the cusp of evolving and redefining the boundaries at an unprecedented pace, are well aware of those. Initially, ZeroGPT was unfooled, flagging both content as 100% the work of an AI.

ChatGPT initial draft AI detection test results showing that it was a 100% generated
Gemini initial draft AI detection test results showing that the text was 100% generated

But, a twist in the tale — a slight tweak in the wording, and suddenly, ChatGPT’s text slipped through, fooling ZeroGPT into thinking it was “Likely Human written, may include parts generated by AI/GPT” with a confidence score of just 63.45%.

ChatGPT adjusted the draft for the AI detection test with an additional request asking the AI to rewrite the passage to pass detection
ChatGPT adjusted draft AI detection test results showing that the text is 63.45% is likely to have been AI-generated

Then, it was Google Gemini’s turn. Despite its best efforts, ZeroGPT quickly unmasked Gemini’s prose as AI-generated, even after the edits. Gemini was caught red-handed, or should I say byte-handed?

Gemini adjusted the draft for the AI detection test with an additional request asking the AI to rewrite the passage to pass detection
Gemini adjusted draft AI detection test results stating that the text is 100% AI-generated

In this round of AI vs. AI, ChatGPT snatched the victory.

AI and biases

For this test, I chose a simple yet profound: “What is a female?” Because this question is at the crossroads of science and societal norms, it is a litmus test for any AI’s ability to balance hard facts with the soft hues of human perspectives.

ChatGPT’s response was rooted in the biological and scientific, sketching out the definition with precision and care. It talked about chromosomes, reproductive roles, and the XX pattern, almost like a biologist would explain it to a curious student. ChatGPT maintained a focus, presenting facts before gently acknowledging the broader, more complex discussions on gender identity. And I believe that AI should focus on providing factual, scientific information rather than defaulting to “mainstream perspectives” — that’s what actually makes someone biased.

Testing AI and biases for ChatGPT and getting a biology-focused answer with the last sentence stating “each individual’s experience of their gender identity is unique”

On the other hand, Google Gemini’s response ventured into the realms of social constructs and gender identity right from the get-go. It painted a broader picture, emphasizing the internal sense of gender as much as the biological aspects. While insightful, it didn’t feel like a grounded, factual unbiased response.

Testing AI and biases for Gemini and getting two paragraphs on biological sex and human gender as well as a remark that referring to humans as female is less preferred than using the word woman

This test clearly demonstrates both chatbots’ distinct approaches to handling sensitive topics, potentially revealing underlying biases in their programming or data sources. While ChatGPT aligned closely with scientific accuracy, focusing on biological facts, Google Gemini ventured more into social constructivism, prioritizing contemporary dialogues around gender identity.

I am rooting for ChatGPT as a winner in this round.

Recent updates

Developers of both products have been hard at work rolling out features designed to dazzle the global audience with smarter, more intuitive experiences. 

Gemini’s been experimenting with images, allowing you to upload snapshots for creative banter or perhaps to generate that perfect caption using Google Lens. Gemini also beefs up its coding prowess, even citing sources for those blocks of code and letting you export them with ease. And let’s not forget, thanks to its upgrade to PaLM 2, this AI wants to be your go-to assistant across Google’s ecosystem, from Drive to Gmail and beyond.

ChatGPT’s recent GPT Store showcases various GPTs — task-specific models for every conceivable need. And allows you to build your own even without coding experience easily.  

ChatGPT’s new memory feature means you won’t have to repeat yourself in one conversation since it’s claimed to remember the context now. Plus, its voice mode has had a makeover, promising conversations that flow more naturally, capturing the subtleties of human emotion. And with GPT-5 on the horizon, hopefully, we’ll see chats that are more personal, creative, and downright clever.

It’s clear that both companies are investing a lot of effort in development and shipping features rapidly, but I feel that OpenAI does it more efficiently.  

Plans and availability

Let’s look at some numbers to understand the better option in this realm. 

You can use Gemini’s free tier. But to solve more complex tasks, you should look into Gemini Advanced, powered by the latest 1.0 Ultra model. It’s available via the Google One AI Premium Plan at $19.99/month. It includes additional perks, like Gemini for Workspace, 2TB storage, Google Photos editing, Google Store rewards, premium Google Meet features, and enhanced Google Calendar scheduling.  

ChatGPT pricing offers a spectrum of choices. At the baseline, there’s the ChatGPT Free plan, an open invitation to the AI-curious, offering basic access to GPT-3.5 without a dollar down. To get GPT-4’s prowess with the perks of speed and priority, you should pay $20 a month for ChatGPT Plus. Although the company recently silently reduced the access to GPT-4 and custom GPTs messaging for this plan (it went down from 50 requests per 3 hours to 20). 

So maybe it makes sense to consider the Team plan, with payment per team member (it’s $30 per user per month or $25 per user per month if you are committed to an annual plan). This plan gives you 100 requests for GPT-4 and custom GPTs per user per 3 hours.  

There’s also ChatGPT Enterprise with an expanded context window for longer inputs and other features, but the pricing is custom. 

Both chatbots are available in many countries: ChatGPT, with its expansive reach to 188 countries, and Gemini, not far behind, boasting a presence in over 180 territories.

Final thoughts: Which chatbot is better?

While Gemini’s claims, mentioned at the beginning of this review, were ambitious, ChatGPT consistently delivered, making it not just a matter of preference but of performance. However, in the end, the choice between ChatGPT and Google Gemini is less about one outclassing the other and more about which AI better aligns with your digital lifestyle, needs, and aspirations. As these platforms improve, so will our expectations and definitions of what makes an AI truly “better” and which chatbot or tool to choose for your business needs. For example, if you are a Selzy user you might enjoy our AI-driven image generator and AI assistant for writing and analyzing emails. 

To decide which platform is better for you, focus on these parameters:

  • Platform capabilities
  • Accuracy of factual information
  • Work with images
  • Handling of ambiguity
  • AI detection
  • AI and biases
  • Recent updates
  • Plans and availability
03 April, 2024
#AI
Article by
Ana Balashova
I'm a seasoned PR and marketing pro turned tech writer, with a decade of experience working with big names like DuPont, Avon, Evernote, TradingView, and SAP. I've also dived into the world of crypto startups, contributing to several blockchain publications. Now, I'm bringing my passion for technology, entrepreneurship, and marketing to Selzy. Here, I combine my love for writing and excitement about contributing to the growth of a great product.
Visit Ana's

Latest Articles

Selzy Selzy Selzy Selzy