Google’s innovative Search Generative Experience (SGE) is about to step up its game. Initially designed to revolutionize the search experience by introducing a more conversational approach, SGE now boasts the capacity to produce images through user prompts, echoing a feature akin to Bing’s integration with OpenAI’s DALLE-E 3. Moreover, SGE now permits users to draft content within its interface, offering the flexibility to modify the length and tone of the generated content.
This remarkable enhancement is not a standalone upgrade. In recent times, Google has been consistently rolling out features for SGE, reflecting the rapid progression in the realm of AI. A few notable updates include SGE’s ability to craft AI-driven summaries, offer explanations for unfamiliar terms, and implement several coding, travel, and product search optimizations.
Diving into the specifics of the image generation function, users can now input particular cues – like specifying a drawing, photo, or painting – and SGE will immediately display four associated results within its conversational interface. Once the results are available, users can either download them directly as .png files or refine the original prompt to yield a fresh batch of images. This sophisticated capability is backed by Google’s proprietary ‘Imagen’ text-to-image model.
Google’s New Feature Interface
![AI Search](https://aifluke.com/wp-content/uploads/2023/10/image-3-1024x396.png)
![AI Search](https://aifluke.com/wp-content/uploads/2023/10/image-3-1024x396.png)
And that’s not all; this image generation utility will also extend to Google’s primary Image Search. If your search doesn’t yield the desired picture, a prompt box will emerge within the results, allowing you to craft a brand-new image on the spot.
However, there’s a catch. Due to concerns surrounding AI-generated inappropriate content, Google is limiting this innovative feature to users who are 18 and above. This decision is a bit surprising, especially since SGE recently became accessible to teenagers aged between 13 and 17 in the U.S.
Google’s SGE Feature
Google’s overarching goal is the responsible dissemination of this technology. To ensure this, they have incorporated stringent filtering measures, as informed to TechCrunch. The aim is to circumvent the generation of images that might be “deceptive, harmful, or explicit,” or those that contradict Google’s set guidelines for generative AI. Additionally, Google has placed restrictions against producing images showcasing realistic faces or referencing notable individuals. These measures aim not only to curb unsuitable content but also to deter the spread of false information.
But there’s an undeniable challenge. As recent incidents with Meta and Bing showcased, people often find innovative ways to exploit AI image generation tools, producing content that is either shocking or inappropriate, despite the safeguards in place.
Google is candid about the imperfections of its tools, and this honesty is evident in its decision to maintain its features as opt-in via Google Search Labs. To ensure that the AI aligns with user expectations, the company has integrated a feedback system within SGE. This will empower users to report any inconsistencies or misuse they encounter. While Google is confident about the effectiveness of its filtering processes, it predicts users might come across more blocks than they anticipate, subtly hinting at the robustness of their safety mechanisms.
Another safeguard in place is the inclusion of metadata within the images, clearly labeling them as AI-generated. Adding another layer of transparency, these images will also carry an embedded watermark that’s invisible to the naked eye. This watermarking technique springs from SynthID technology, a joint venture unveiled by Google Cloud and Google DeepMind in late August.
Google’s SGE Interface
![AI Search](https://aifluke.com/wp-content/uploads/2023/10/image-2-1024x458.png)
![AI Search](https://aifluke.com/wp-content/uploads/2023/10/image-2-1024x458.png)
Venturing into the realm of content drafting, SGE isn’t a novice. Prior to these updates, it already showcased capabilities as a writing assistant. Users could generate written drafts — a feature integrated with the notion that users might wish to pen down emails to businesses discovered via Google Search. However, Google has now expanded this utility. The enhanced draft-writing feature can produce varied content — be it lengthier or concise, or modulating the tone between casual and formal.
SGE’s Writing and Image Features
The new inclusions in SGE are not just about creation but also about seamless sharing and storage. Users can effortlessly export the written drafts to Google Workspace tools like Gmail or Google Docs. Simultaneously, any images they craft can be directly stored in Google Drive.
The anticipation around these features is palpable, and the wait won’t be long. These functionalities will start rolling out tomorrow for a subset of SGE users. Over the following weeks, a broader audience will have access. Initially, they will be accessible for users who have activated SGE through Google Search Labs and will cater only to English speakers in the U.S. This might seem restrictive, given that SGE recently made its debut in India and Japan. However, global expansion seems to be on the horizon.