TL;DR
- Sources claim that Google has rescheduled unannounced launch events for the company’s new LLM.
- The delay may have been due to unreliable responses to some non-English prompts.
- Google may be planning to do a virtual preview of the new AI this week instead.
Google has been working on a new large language model (LLM) called Gemini that will eventually substitute its current PaLM 2 AI architecture. It appears Google may have initially planned to launch the next-gen LLM this month, but the launch was reportedly delayed. However, Google may now be planning to do a virtual preview of Gemini instead.
According to a report from The Information, sources with knowledge of the situation claim that there were unannounced Gemini launch events scheduled for this week. Those launch events — which would’ve taken place in New York, Washington, and California — are said to have been rescheduled for January 2024.
It appears the reason for the delay may be due to a reliability issue. Reportedly, Gemini’s reliability is a bit shaky when submitting prompts or inquiries that aren’t in English.
While the in-person launch events may have been delayed, we may still get the chance to see what Google’s next-gen AI is capable of. The outlet’s unnamed source says that the firm is now planning to do a virtual preview that could happen as soon as this week.
Gemini is Google’s most powerful AI architecture to date, and the company hopes it will match or even exceed OpenAI’s GPT-4. Its main differentiator is that it is multimodal and can be trained on more than just text. Gemini can reportedly create text, images, audio, videos, and other data types.
