Google's demonstration of its new AI model, Gemini, initially received high praise, but has since faced accusations of being "basically entirely fake."
Oriol Vinyals from Google Deepmind clarified that the demo was edited for brevity, revealing that Gemini's interactions were text-based, not voiced, and significantly slower than depicted.
Google had included a disclaimer in the YouTube video stating that latency was reduced and outputs were shortened for brevity, but this didn't prevent a wave of criticism on social media.
Critics, including software developers and even Google employees, have expressed that the demo misrepresented Gemini's actual capabilities.