Google Gemini vs ChatGPT: 3 mins recap on key differences, similarities, and better of the two.

Uncover key differences, capabilities, and future implications in this concise, expert analysis of the two AI giants

The launch of Gemini has created a BUZZ 🐝 worldwide, and there are multiple discussions on this next big AI. How does the super rookie Google Gemini compare to the OG ChatGPT? Today, let’s discuss what we know about them.

In today’s rundown:

  •  GPT-4 vs. Gemini: Key Similarities

  • GPT-4 vs. Gemini: Key Differences

  • 🤔 GPT-4 vs. Gemini: Is the later the better?

  • 😟 Should you be concerned about the launch of Gemini?

    read time: 3 mins

Key Similarities

GPT-4 and Gemini are both Large Language Models capable of understanding and generating responses across various mediums.

At a high level, they are similar.

Currently, GPT-4 is OpenAI's largest model, comparable to Gemini Ultra in terms of parameters and functionality. The free version of ChatGPT is based on ChatGPT-3.5. GPT-3.5 is similar to Gemini Pro, which is used in Google's chatbot, Bard.

Gemini offers different sizes:

  • Gemini Ultra: Full-size model for complex tasks.

  • Gemini Pro: Balanced performance and scalability, integrated into Bard.

  • Gemini Nano: Smallest on-device models available in two sizes:

    • Nano-1: 1.8B parameters

    • Nano-2: 3.5B parameters

Key differences

Gemini is advertised for surpassing human experts in terms of accuracy. Google's release notes highlight the Massive Multitask Language Understanding (MMLU) benchmark, where Gemini Ultra achieves a benchmark of 90% accuracy compared to the average human expert's 86%. Gemini also outperforms GPT-4 on seven out of eight text benchmarks, with the only exception being commonsense reasoning and problem-solving (HellaSwag). For more details, refer to the comparison below.

The values included in this figure were obtained from Google Deepmind's own calculations.

Despite benchmarking higher, Gemini's wins seem relatively minor. The models are separated by only a few percentage points on each test.

Read more about the tests here.

Is Gemini more multimodal than ChatGPT?

In the field of Generative AI, "multimodality" refers to a model's ability to process and generate data in various formats. Traditionally, AI models were limited to working with text only, but now they can handle code, images, video, and audio.

While ChatGPT has incorporated multimodal prompts and can generate images with Dall-E, Gemini seems to have taken it further. Based on the demo, Gemini appears to have more advanced multimodal prompting and response capabilities. However, is this the case?

Google was called out for the “fake” demo video?

The demo video released by Google showcased Gemini's impressive ability to reason, understand context, and respond in various media formats. The video appeared to be a live feed, which made Gemini's low latency capabilities seem promising for real-time AI assistance.

However, upon reading the description of the YouTube video, Google provided a brief disclaimer:

"For the purposes of this demo, latency has been reduced and Gemini outputs have been shortened for brevity."

Therefore, the live camera feed was not live. Instead, Google prompted Gemini to use a series of still images and combine them with text. Gemini responded effectively to these visual inputs, not through a live camera feed.

You can watch the video here.

So tell me which one is better, Google or OpenAI?

Instead of speculation and assumed capabilities, I created an article to answer this question, backed with data and scientific tests. You can read about it here.

Author’s note: Should you be concerned about the launch of Gemini?

Gemini appears to be a strong competitor to ChatGPT and GPT-4, but it hasn't made significant breakthroughs with Gen AI. Nonetheless, its launch demonstrates the expansion of AI and the drive for innovations.

In a world with emerging AI-powered apps, what does this mean for blue-collar engineers concerned about AI replacing professionals?

While Gen AI can enhance productivity in various fields, I have reservations about completely replacing skilled professionals. Competent professionals are hired for their problem-solving abilities, not just as copy machines. As a developer, have you reviewed the ChatGPT code? Sometimes it doesn't function as expected and produces flawed output. Are you working in a specialized field? Gen AI heavily relies on publicly available information, so it may not provide valuable insights specific to your niche. These are some limitations of AI for professionals.

Most individuals, including professionals, may never directly interact with an LLM model. However, leveraging its capabilities to enhance workflows and productivity will be necessary. This is where AI excels: in automating and streamlining processes.

Easier said than done. There is so much information about AI. How can I leverage AI for myself 😕 ? Our newsletter focuses on the latest and most useful AIs used and chosen without overselling their capabilities. Please stay connected if you want a good daily read to stay updated with AI tools and trends. ❤️ 

Which AI do you think will be better? 💞

Cast your vote to win a $20 gift card and contribute toward your favorite tools. Your selection will guide others to discover the emerging useful tool, and the top pick will be honored as the AI overlord for the month.

Which one is the better AI?

Login or Subscribe to participate in polls.

That’s a wrap! 🌯

Thank you for reading ❤️I hope you will find these insights useful. Please contact us at [email protected] if you have suggestions, feedback, or anything!