Google Rolls Out Imagen 3: Enhanced Image Generation in Gemini After Past Controversy

Update: 2024-08-29 11:04 IST

Google has unveiled new enhancements to its Gemini platform, including the much-anticipated Imagen 3 and a feature called Gems. These updates are being gradually introduced, starting with subscribers to Gemini Advanced.

Earlier this year, Google paused its AI-driven image generation tool following a wave of criticism directed at the previous version, Imagen. The tool had sparked backlash due to concerns over biased outputs and inaccurate depictions of historical scenes, particularly regarding racial representation. This prompted Google to halt the feature and focus on addressing these issues.

ADVERTISEMENT

In February, the controversy surrounding Google's image-generation tool gained significant attention. Critics pointed out that the tool's outputs often contained inaccuracies, leading to concerns about the potential societal impact. For instance, it was noted that the tool sometimes produced racially insensitive images or misrepresented historical events. As a result, Google made the decision to pause the tool's availability and commit to making necessary improvements.

Now, six months later, Google has reintroduced its image generation capability with Imagen 3, an improved version of the previous tool. According to Dave Citron, the Senior Director of Gemini, the new version offers enhanced creative image generation and will soon be accessible to all users. Not only has the creative functionality been refined, but the tool's availability has also been expanded to support users in various languages.

In a recent blog post, Citron stated, "Imagen 3 sets a new standard for image quality, generating images with just a few words." He emphasized that the company has made "significant progress" in refining how the tool generates images of people.

"We’ve worked to make technical improvements to the product, as well as improved evaluation sets, red-teaming exercises and clear product principles," Citron explained. Red teaming, which involves testing a product by simulating real-world scenarios, was employed to identify potential vulnerabilities before the tool's release. While Citron acknowledged that Imagen 3 is not without its flaws, he assured users that Google is eager to receive feedback and make further improvements as needed.

Citron also clarified that users would not be able to generate photorealistic imagery, depictions of identifiable individuals, portrayals of minors, or any excessively violent, gory, or sexual content using Imagen 3. The rollout of this feature will be phased, eventually reaching a wider audience.

In addition to Imagen 3, Google has also introduced a new feature called Gems. Over the coming days, subscribers to Gemini Advanced, Business, and Enterprise can create and interact with Gems, a custom version of Gemini that was first previewed at the I/O conference. With this feature, users can customize Gems to act as experts on specific topics or tailor them to meet particular goals.

The Google said, "With Gems, you can create a team of experts to help you think through a challenging project, brainstorm ideas for an upcoming event, or write the perfect caption for a social media post. Your Gem can also remember a detailed set of instructions to help you save time on tedious, repetitive or difficult tasks." The rollout of Gems will be available on both desktop and mobile devices to Gemini Advanced, Business, and Enterprise users in more than 150 countries, supporting most languages.

n
ADVERTISEMENT

Tags:    
ADVERTISEMENT

Similar News