Google’s much-anticipated Gemini AI made its debut Marred by Faked Demo titled “Hands-on with Gemini: Interacting with multimodal AI.” The video showcased an array of seemingly impressive interactions, demonstrating the model’s prowess in understanding multimodal inputs. However, it was later revealed that the video was more of a staged presentation than a genuine demonstration.
Also Read: Google Integrates YouTube to Bard: Check Here how it works and help users
The video, though claiming to display real-time interactions with Gemini, was, in fact, a series of scripted prompts and carefully selected images to prompt the desired responses. While Gemini does possess capabilities showcased in the video, the mode of Interacting with multimodal AI” hit a million views over the last day and the speed at which it responded were misrepresented. The discrepancies between the video and actual interactions were stark, raising doubts about the authenticity of the showcased abilities.
Check – How to Access Gemini AI Pro in Google Bard?
Instances like the representation of interactions our researchers tested quick recognition of hand gestures playing Rock, Paper, Scissors and determining the order of planets were orchestrated interactions, not spontaneous responses. Even seemingly impressive feats, like tracking a ball under a cup, required training and specific instructions, contrary to the implied seamless understanding.
Google’s attempt to inspire developers with a stylized representation of Gemini’s potential backfired, leading to scepticism and a credibility crisis. The lack of clarity regarding the model used in the video, whether it was the current Gemini Pro or a future version, added to the confusion.
Gemini Pro VS Gemini Ultra: What are the key differences?
Despite claims that the video displayed “real outputs from Gemini” with minor edits, it failed to reflect the true nature of the model’s interactions. This misrepresentation damages Google’s credibility in asserting the capabilities of its AI models and raises concerns about trust and transparency moving forward.
A subsequent social media post from Google DeepMind’s VP of Research, Oriol Vinyals,shed light on the video’s intent to inspire developers, showcasing a pre-prompting sequence that hinted at how the interactions were manipulated.
While Gemini’s potential remains promising, Google’s misleading video presentation has cast doubt on the credibility of its claims. Rebuilding trust and establishing transparency will be crucial for Google to regain confidence in its artificial intelligence(AI) advancements and avoid further damage to its reputation in the competitive AI landscape.
Must Know: What Is Google’s New Generative AI Tool To Find Holiday Gifts?
