Google has made some progress on its artificial intelligence service Gemini.Many updates, which provides a lightweight version of the model designed for the free version of the serviceGemini 1.5 FlashAt the same time, adjustments are also made to address the "hallucinations" that are prone to occur in artificial intelligence interactions, so that it can answer questions that are more in line with the facts during interactions.
Gemini 1.5 Flash was announced during Google I/O 2024 this year. It can also generate, summarize, interact, or analyze data from large documents and tables. Compared with the Pro version, it can operate in a lighter form and can also operate in multi-mode. It also supports 100 million sets of word inputs and can analyze documents up to 1500 pages or more than 3 lines of program code. It can also analyze videos up to 1 hour long, but in terms of functionality, it can basically be regarded as a simplified version of Gemini 1.5 Pro.
As for the previous update, Google upgraded Gemini 1.5 Pro to support 200 million word inputs, and can process documents of more than 140 million words, more than 6 lines of program code, and up to 2 hours of video content. It is expected to be upgraded before the end of this year.
This update for the free service not only replaces the lightweight Gemini 1.5 Flash, but also increases the length of the system's context to 32K characters, which is about four times larger than the previous version, allowing free service users to interact with longer content.
The update to the free service also includes the ability to import files directly from Google Drive or other devices to the Gemini service, allowing the Gemini service to analyze archive files more quickly. However, there will still be a 32K word length limit, and by switching to Gemini 1.5 Flash, the language content support has been increased to 40, and it will currently be available in 230 countries and regions around the world.
In addition, content generated through the Gemini service will also be accompanied by reference sources. At the same time, the artificial intelligence-generated content will be compared through the Double Check function to see if it is consistent with the facts. The matching content will be marked in green, and the different content will be marked in orange. However, if insufficient information is found for analysis, or if the generated content itself does not have a standard answer, the marked content will not be displayed as a reference.
Google has also expanded its Gemini service to include teens. While users in many countries and regions can use it as long as they are 13 years old or older, some European regions require users to be 16 years old or older to comply with local laws. Regarding this policy adjustment, Google stated that it has collaborated with multiple child safety and development experts to ensure that the service does not affect the mental development of young people and that it can unlock greater creativity through artificial intelligence.
