Google is making changes to the free version of Gemini, its AI-powered robot, so it can keep up with competitors like Anthropic and OpenAI. The updates are meant to make the platform work better and be open to more people.
Google stated in May that Gemini 1.5 Flash would be a lightweight multimodal model that would be available on the web and on phones starting Thursday in 40 languages and about 230 countries. Google says that Gemini 1.5 Flash improves quality and latency, with gains that are especially clear in reasoning and understanding images.
Besides being better for Google, running on the back end might also cost less.
When Google showed off Gemini 1.5 Flash, the company said it was a “distilled” and very efficient form of Gemini 1.5 Pro, made for what they called “narrow,” “high-frequency” generative AI workloads. Given how much it costs to run a chatbot platform like Gemini (see OpenAI’s ChatGPT bills), Google is probably eager to take advantage of any chance to save money, especially if those chances also improve performance in other areas.
Google says that Gemini’s context window will grow to 32,000 tokens, which is about 24,000 words (or 48 pages of text), in addition to the new base model.
The input data (like text) that a model looks at before making output (like more text) is called context, or context window. Some benefits of models with bigger contexts are that they can recap and reason over longer chunks of text and files (in theory), and in the context of a chatbot, they’re less likely to forget about things that were just talked about.
Before, you needed Gemini Advanced, the paid version of Gemini that was only available to people with Google’s $20-a-month Google One AI Premium Plan in order to send files to Gemini for analysis. To make things better, Google says it will soon let all Gemini users add files from Google Drive and local devices.
As VP of engineering at Google, Amar Subramanya wrote in a blog post that TechCrunch saw, “You’ll be able to do things like upload your economics study guide and ask Gemini to make practice questions.” “Gemini will soon also be able to look through your data files and find insights that you can show off in charts and graphs.”
Google is testing a feature that puts links to related web material below some Gemini-generated answers. This is meant to help stop hallucinations, which happen when a generative AI model like Gemini 1.5 Flash makes things up. English-language In some areas, Gemini users will see a “chip” icon at the end of a Gemini-written paragraph. This icon will lead to websites (or emails, if you’ve allowed Gemini to view your Gmail inbox) where you can learn more.
This comes after it was found that Google’s generative AI models often have very bad hallucinations, like putting nontoxic glue in a pizza recipe and making up fake book reviews that were attributed to real people. Google added a “double check” tool to Gemini earlier this year. This feature shows which statements from Gemini are supported or disagreed with by other online sources. But the related material links look like an attempt to make it clearer where Gemini might be getting its information.
This reporter wants to know how often and correctly Gemini will find similar links. TBD.
That being said, Google isn’t waiting to fill the channels.
Google first released Gemini in Messages earlier this year for a small group of devices. Now, the company is rolling it out to the European Economic Area (EEA), the UK, and Switzerland. People will be able to chat in new languages like French, Polish, and Spanish. To bring up Gemini in Messages, users must tap the “Start chat” button and choose Gemini as a chat partner.
Google is also releasing the Gemini mobile app in more countries and making Gemini available to more teens around the world.
In June, the company released Gemini, a service for teens that lets them use their school accounts to sign up, but not in all countries. Next week, that will change because Gemini will be open to teens in all the same countries and areas where it is usually open to adults.
Google says it’s adding “additional policies and safeguards” to protect teens at the same time as the rollout, but it doesn’t say what those are. Google calls it a “AI literacy guide” to “help teens use AI responsibly.” There is also a new onboarding process that is made just for teens.
Also Read: Gemini Isn’t as Good at Studying Data as Google Says It is
Lots of people have different ideas about whether kids are using generative AI tools the right way or abusing them. Google is probably trying to avoid headlines that say Gemini can plagiarize essays or give kids bad advice about their personal problems. To do this, it is doing everything it can to stop the worst from happening.
What do you say about this story? Visit Parhlo World For more.
