Google Unveils New and Improved AI Model: Gemini 1.5 Pro

Google, a leading technology company, recently unveiled its latest advancement in artificial intelligence (AI) technology: Gemini 1.5 Pro. This new model is described as delivering “dramatically enhanced performance” over the previous version, Gemini 1.0 Ultra. The unveiling of Gemini 1.5 Pro follows the rebranding of Google’s Bard chatbot to Gemini, aligning with the new model’s more powerful and versatile capabilities.

Google’s CEO, Sundar Pichai, and Google DeepMind CEO, Demis Hassabis, emphasized the importance of ethical AI safety while touting the new model’s rapidly advancing capabilities. The company’s need to assure AI skeptics and government regulators of ethical AI safety is evident, as well as the need to stress the models’ accelerating performance for potential customers and investors concerned about the company’s reaction to OpenAI’s breakout success with ChatGPT.

According to the announcement blog post, Gemini 1.5 Pro delivers comparable results to Gemini 1.0 Ultra but does so more efficiently, with reduced computational requirements. The multimodal capabilities include processing text, images, videos, audio, or code. As AI models advance, they will continue to offer a more versatile array of capabilities in one prompt box.

Google’s Gemini 1.5 Pro can handle up to one million tokens, or the units of data AI models can process in a single request. The company claims that Gemini 1.5 Pro can process over 700,000 words, an hour of video, 11 hours of audio, and codebases with over 30,000 lines of code. A version that supports up to 10 million tokens has even been “successfully tested.”

Gemini 1.5 Pro maintains high accuracy in queries with larger token counts when it has more new data to learn. It impressed in the Needle In a Haystack evaluation, where developers insert a small piece of information inside a long text block to see if the AI model can pick it out. Gemini 1.5 Pro could find the embedded text 99 percent of the time in data blocks as long as one million tokens.

The company says Gemini 1.5 Pro can reason about various details from the 402-page Apollo 11 moon mission transcripts and analyze plot points and events from an uploaded 44-minute silent film starring Buster Keaton. Google is continuously developing new evaluations and benchmarks for testing its novel capabilities.

Google is launching Gemini 1.5 Pro with 128,000-token capabilities, the same number at which OpenAI’s publicly announced GPT-4 models max out. The company plans to introduce new pricing tiers that support up to one million-token queries.

Gemini 1.5 Pro is adept at learning new skills from information in long prompts without additional fine-tuning (“in-context learning”). In a benchmark called Machine Translation from One Book, the model learned a grammar manual for Kalamang, a language with fewer than 200 speakers globally that it hadn’t previously been trained on. The company claims that Gemini 1.5 Pro learned to perform at a similar level as a human learning the same content when translating English to Kalamang.

In a significant development for developers, Google says Gemini 1.5 Pro can perform problem-solving tasks across longer code blocks. When given a prompt with more than 100,000 lines of code, it can better reason across examples, suggest helpful modifications, and give explanations about how different parts of the code work.

On the ethics and safety front, Google is taking the same approach to responsible deployment as it did with Gemini 1.0 models. This includes developing and applying red-teaming techniques, where a group of ethical developers essentially serve as devil’s advocate, testing for “a range of potential harms.” The company also heavily scrutinizes areas like content safety and representational harms and continues to develop new ethical and safety tests for its AI tools.

Google is launching Gemini 1.5 in early access for developers and enterprise customers, with plans to make it more widely available eventually. Gemini 1.0 is currently available for consumers, alongside a Pro variant that costs $20 monthly.

In conclusion, Google’s Gemini 1.5 Pro represents a significant leap forward in AI technology, offering enhanced performance, multimodal capabilities, and the ability to learn new skills from long prompts. The company’s commitment to ethical AI safety and responsible deployment is also noteworthy. As AI continues to evolve and become more integrated into our daily lives, advancements like Gemini 1.5 Pro will play a crucial role in shaping the future of technology.