Toolhacker logo
Innovative 🚀 | Solutions 💡 | Empowering 💪 | You 🙋
Gemini Advanced
Apps

Gemini Advanced

Sep 14, 2024

In February, Google released a significant update to its Bard chatbot, called Gemini. Gemini included a new mobile app and access to its eagerly awaited Ultra model. Today, its premium version, Gemini Advanced, provides access to Ultra 1.0, Google’s “largest and most capable” AI model that reportedly competes with GPT-4 on performance benchmarks.

This latest underlying LLM, which Google reports is still Gemini 1.5 Pro, continues to present many challenges. For example, when confronted with an error about a Ford Fiesta question, Gemini keeps repeating the same erroneous answer from a previous question:

“The window sticker you provided doesn’t contain any information about the underlying Large Language Model (LLM) used by Gemini Advanced. Information about the specific LLM used by Gemini Advanced is typically not included in documents like window stickers, which focus on the vehicle’s specifications and features.”— Gemini Advanced
14-Sep-24

We also noticed that Gemini Advanced is trained to refrain from continuing threads it mistakenly believes are politically tainted, like when we asked it to provide background on investments made by several billionaires reportedly tied to a particular candidate.

When asked to write about a specific topic creatively, it did produce an appealing short story, putting it on par with ChatGPT. However, the story veered too much in the direction of fiction, which is not what we had intended. However, that might have been remedied with prompt engineering.

Gemini Advanced is competitively priced with ChatGPT Plus at $20/mo., although Google includes the Advanced version as part of its newly rebranded Google One AI Premium (2 TB) membership plan. That move is clearly intended as a preemptive strike against OpenAI.

A Gemini mobile app for Android and iOS offers improved on-device help with voice, text, and picture capabilities. Using a technology called “SynthID,” images created with Gemini will now have an undetectable digital stamp.

Gemini Advanced excels when collaborating with Google products. Thanks to integration with Google Maps, it provides improved information on current events and businesses. However, its image generation capabilities fall short compared to models like DALL-E 3.

  • Advanced tasks – Gemini Advanced handles tasks such as translation, handling multiple instructions, and image generation.
  • Summaries – It accurately summarizes complex academic papers.
  • Google Maps integration – A Google Maps integration provides better information about businesses compared to ChatGPT Plus.

The primary advantages of Gemini Advanced over ChatGPT Plus are:

  • Better integration – Gemini Advanced can access information from Google Search, Google Maps, and other Google products, which gives it a significant advantage over ChatGPT Plus when it comes to tasks such as finding information about businesses, getting directions, or making reservations.
  • More accurate information – Gemini Advanced is trained on a larger dataset than ChatGPT Plus, which gives it access to more up-to-date and accurate information.
  • More complex tasks – Gemini Advanced can handle more complex tasks than ChatGPT Plus, such as translating text, generating images from text prompts, and writing creative content.

Finally, on Feb. 15, 2024, Google introduced Gemini 1.5, which delivers dramatically enhanced performance and a new Mixture-of-Experts (MoE) architecture. It also introduces a breakthrough experimental feature in long-context understanding.

Gemini 1.5 Pro comes with a standard 128,000 token context window. But a limited group of developers and enterprise customers can try it with a context window of up to 1 million tokens to process vastly more data than previous AI systems. Gemini 1.5 Pro can handle 700,000 words of text, 30,000 lines of code, 11 hours of audio, or 1 hour of video — enabling groundbreaking processing capabilities.

As we predicted, artificial intelligence will only accelerate innovation, and we are already starting to see the results of that acceleration. 🤯

Michael Tchong

Michael Tchong

Founder, Author, Adjunct Professor, Futurist

Michael is the founder of Toolhacker LLC, an innovation speaker and adjunct professor at the University of San Francisco, futurist, and author of “Ubertrends — How Trends And Innovation Are Transforming Our Future.” He is also the founder of four ahead-of-the-wave startups, including MacWEEK, Atelier Systems, CyberAtlas, and ICONOCAST.

Sign Up Today

And get your free copy of Generative AI

[hubspot type="form" portal="22379253" id="b966525b-7305-4766-bc05-1ecb0d1f1017"]