What the Gemini Image Generation Fiasco Tells Us About Google’s Approach to AI

What the Gemini Image Generation Fiasco Tells Us About Google’s Approach to AI - Entertainment - News

Google’s ai Controversies: From Sentient Claims to Image Generation Fiasco

In July 2022, before the release of ChatGPT, one of Google’s engineers claimed that their LaMDA ai model had become sentient. Google responded by stating their commitment to responsible innovation and the seriousness with which they take ai development.

Google’s Cautious Approach to ai: The Gemini Image Generation Incident

Fast forward to the Gemini image generation fiasco, where a user requested an image of “America’s Founding Father.” The model responded with images of diverse individuals, excluding white Americans. Critics accused Google of anti-White bias and capitulating to “wokeness,” leading Google to temporarily disable the image generation feature.

Google’s Diversity Focus and Model Failure

Google explained the issue as a result of failed tuning for cases where a range was not intended. The model, which had been optimized to show diverse ethnicities, lacked ethical frameworks and rigorous training for different contexts.

Optimization and Ethical Concerns

Many ai companies, including Google, optimize their models using data scraped from the internet. This data often contains discriminatory language and racist overtones, leading to biased results. Margaret Mitchell, Hugging Face’s Chief ai Ethics Scientist, suggested that this might be due to under-the-hood Website image optimization and a lack of clear ethical guidelines.

Gemini’s Diversity-Specific Prompting

To address this, Gemini may use diversity-specific prompts to generate diverse results. However, these universal prompts could lead to unintended consequences, such as misrepresentations or exclusions.

Gemini’s Text Generation Model and Over-Caution

Besides the image generation issues, Gemini’s text generation model refuses to answer certain sensitive prompts, leading to unreasonable behavior. For instance, it fails to call out absurdities or acknowledge that pedophilia is wrong.

Google’s Timid Culture and Future Plans

Ben Thompson described Google as having become timid, sacrificing its mission to “organize the world’s information and make it universally accessible and useful.” At Google I/O 2023, the company announced a “bold and responsible” approach to ai. However, critics argue that this timid culture has only worsened Google’s situation.


Google’s overly cautious approach to ai, shaped by its culture and the increasingly polarizing world, has led to controversies. From claims of sentience to image generation fiascos, the consequences of this approach can be far-reaching.