Google delays launch of Gemini, its answer to GPT-4

We'll have to wait a little longer before we see Gemini, Google's next AI model that aims to compete with GPT-4, in action. According to The information, the company has postponed a series of events where it was planned to publicly introduce the new technology. The Gemini presentation was supposed to be held this week in Washington, New York and California, but the corporation chose to postpone them until January 2024 in order to have time to eliminate some technical shortcomings.

Google will need more time to work on Gemini

In the case of Gemini, Google chose to devote a few more weeks of work to it. Mountain View officials have been committed to having the new model introduced since its first release as an alternative that matches or extends the capabilities of GPT-4. One of the key points in achieving this goal is the built-in support for multiple languages. And this is where, presumably, the biggest problem lies.

According to The Information's sources, Google discovered that Gemini would not properly process entered commands if they were not written in English.

As for Gemini and its capabilities, the details provided by Google are really sparse. The company announced its development at an event hosted by Google DeepMind, a result of the merger of Google Brain and DeepMind. The only thing that was mentioned about it is that it is a latest generation multimodal AI model.

The existence of Gemini is curious given that Google already has PaLM 2. In fact, it is this model that powers Bard, a chatbot that competes with ChatGPT. Although it has been used in other fields, for example in medicine with Med-PaLM 2.

It is expected that, as with PaLM 2, Google will offer Gemini to developers in different sized versions. It is not yet known how the company plans to integrate it with its current products or services.