Google Building AI System to Outperform ChatGPT, Says Google DeepMind CEO
According to Demis Hassabis, the CEO of Google DeepMind, Google is currently developing an AI system called “Gemini” that will surpass the capabilities of ChatGPT. This new system is expected to take several months to complete and could cost the company hundreds of millions of dollars. Gemini will be focused on working with text and will share many similarities with ChatGPT-4.
However, Google’s goal is not just to replicate existing AI models. The company’s engineers are incorporating techniques from AlphaGo, an AI system that famously defeated a champion in the board game Go. By integrating AlphaGo’s planning and problem-solving abilities into Gemini, Google aims to enhance the capabilities of this new AI system. The large language model capabilities will be combined with these skills to create an advanced AI system.
AlphaGo was developed using reinforcement learning, where software repeatedly attempts to solve a task and adjusts its performance based on feedback. Additionally, the Gemini project may draw ideas from other areas of AI, such as robotics and neuroscience, to further enhance its capabilities.
The announcement of Gemini took place during Google’s I/O conference. Google stated that Gemini is designed to be multimodal, with efficient tool and API integrations, and built to enable future innovations such as memory and planning. While it is still an early-stage project, Gemini has already demonstrated impressive multimodal capabilities not seen in previous models. The system will be available in various sizes and capabilities.
In April, a leaked internal document from Google raised concerns about the company losing its competitive edge in AI to the open-source community. The document mentioned the community’s ability to quickly adapt and make adjustments to their AI projects. As a response, Google removed the waiting list for its chatbot Bard, a competitor to Open AI’s ChatGPT. However, the European launch of Bard had to be delayed due to privacy concerns.
Considering the negative impact of Bard’s factual mistake during its initial demo, which caused Alphabet, Google’s parent company, to lose $100 billion in market value, it is likely that Google will take extra precautions in designing Gemini to be highly reliable and error-free.
Editor Notes
In an exciting update from the AI world, Google DeepMind’s CEO, Demis Hassabis, has revealed that Google is developing an AI system called Gemini that is expected to outperform ChatGPT. This development signifies Google’s commitment to pushing the boundaries of AI technology. Gemini, which draws inspiration from AlphaGo, will not only have powerful language model capabilities but also planning and problem-solving abilities. Hassabis also hinted at potential innovations in Gemini that will make it even more interesting.
At GPT News Room, we are always thrilled to hear about advancements in AI, especially when it comes from big players like Google. The field of AI continues to evolve at a rapid pace, and breakthroughs like Gemini are paving the way for exciting possibilities. To stay updated with the latest news and developments in AI, make sure to visit GPT News Room.
from GPT News Room https://ift.tt/5ZVBAc6
No comments:
Post a Comment