Google Bard has been talking about shifting to a stronger and more efficient LLM model, Gemini, which intends to challenge the current AI leader, ChatGPT.
ChatGPT has been ruling the generative AI world till now. But an American tech giant, Google, has worked for nearly a year on its state-of-the-art Large Language Model Gemini. It has currently rolled out the Gemini powered capabilities on its in-house chatbot, Google Bard. Considering how incompetent Google Bard’s performance has been compared to ChatGPT till now, the new introduction has much to prove.
Google’s advanced AI model claims to bring better reasoning, faster planning, easier understanding, and many more capabilities that can take over other AI models in the market. However, the release process Google has chosen seems to be quite offbeat. Gemini is the biggest upgrade introduced by Google’s Deepmind team in over a year.
“This is the biggest single quality improvement of Bard since we’ve launched,” said Sissie Hsiao, VP and GM of Assistant and Bard at Google, when introducing the Bard upgrade in a press briefing.
It will be launched in three sizes, Ultra, Pro, and Nano, making the new AI usable on a wide range of devices and platforms. People with the most advanced systems and low-end mobile devices can now access the technology to its full extent. The rollout will be commenced in two phases. First, Gemini Pro will be introduced to power the Bard chatbot and will be limited to text generation capabilities. This version will be available in English in over 170 countries and will be released in more languages and regions soon. After the first phase, Gemini Ultra will be rolled out with enhanced capabilities that include image, video, code, and more.
Before the official rollout, Gemini Pro was put through multiple stringent industry benchmark tests. In 6 out of 8 benchmarks, Gemini Pro outperformed ChatGPT-3.5. It also outperformed ChatGPT on GSM8K, a measurement for analyzing grade school math reasoning. ChatGPT is over a year older than Gemini, and it makes sense to give the benefit of the doubt to Gemini being a new member. While some may think that Gemini is merely catching up, it has introduced itself with the capabilities that outperform ChatGPT-3.5 which is available publicly. This essentially gives Gemini an year to improve much further than GPT would.
“Now with Gemini, we’re one step closer to bringing you the best AI collaborator in the world,” Hsiao noted. That at least seems more honest, as it’s an admission that Bard is not quite there yet.
The power of Gemini will be reflected in the difference of Google Bard’s performance, and that is the only way to access its capabilities. The next generations of Gemini, including the Ultra version will be available on Bard advanced, soon to be released. Gemini Ultra will be able to generate high-quality code in popular languages and also bring a deeper understanding of audio and video content.
The overall Gemini framework is still in its learning phase as the team’s researchers found it to be misaligning answers in multiple native languages, a few months earlier. With this update, we can now finally see an alternative option to ChatGPT with better access to the internet and direct access to one of the largest databases in the world, Google.
Arpit is a dreamer, wanderer, and a tech nerd who loves to jot down tech musings and updates. With a logician mind, he is always chasing sunrises and tech advancements while secretly preparing for the robot uprising.