Google removes Gemini AI demo after Ad Watchdog inquiry

Google has quietly removed a demo video of its Gemini AI, which had garnered attention for its seemingly advanced conversational capabilities.

The move to remove the Google Gemini AI demo comes after the National Advertising Division (NAD) of BBB National Programs, an industry watchdog, raised concerns about the accuracy of the video in depicting Gemini’s true performance.

The video, originally posted by Google in December, showcased Gemini responding quickly to various spoken prompts, including identifying parts of drawings and creating a geography game on the fly. However, the NAD questioned whether the video “accurately depicts the performance of Gemini in responding to user voice and video prompts.”

In response, Google chose to discontinue the promotion of the video rather than engage in a prolonged inquiry. The video contained disclaimers that may have gone unnoticed by many viewers, stating that “latency has been reduced and Gemini outputs have been shortened for brevity.” Additionally, a note near the beginning of the video mentioned, “Sequences shortened throughout.”

Oriol Vinyals of Google DeepMind further clarified that the video was meant to illustrate what “the multimodal user experiences built with Gemini could look like,” rather than represent real-time performance.

As part of its response, Google unlisted the video but has kept a related blog post detailing how the demo was created. Google has not issued an immediate comment on the matter, but in a statement included in NAD’s press release, the company expressed that it is “pleased to accept NAD’s resolution of this matter.”

The incident highlights the ongoing scrutiny of AI capabilities and the importance of transparent representation in promotional materials.