8th December 2023
Google has raised the AI stakes with the introduction of Gemini, an artificial intelligence (AI) model claiming to redefine problem-solving through advanced reasoning capabilities. Tested across 57 subject areas, including mathematics and humanities, Gemini is positioned as Google's "most capable" AI model to date, with the potential to outperform human experts in various intelligence tests.
Unlike its predecessor, the AI chatbot Bard, Gemini is described as a foundational model, not a standalone product, and is poised to seamlessly integrate into existing Google tools, including search and Bard itself.
The model's multimodal capabilities mean it can generalize and seamlessly understand, operate across, and combine different types of information including text, code, audio, image, and video potentially paving the way for transformative innovations in generative AI. Demis Hassabis, the CEO of Google DeepMind, hailed Gemini as a "significant milestone in the development of AI."
Gemini comprises three capability levels: Ultra, Pro, and Nano. The Ultra version, designed for highly complex tasks, purportedly outperforms OpenAI's GPT-4 in 30 of 32 academic benchmarks, although OpenAI has plans to release a more powerful version in the coming year. Gemini Ultra is scheduled for public release early next year.
Gemini Pro is already being integrated into Bard in more than 170 countries including the US but has not yet been approved by regulators in the UK and EU. Google said the upgrade will make Bard "more capable at things like understanding and summarising, reasoning, brainstorming, writing, and planning".
Finally, the Nano version is tailored for local device use, including Google's Pixel smartphones, facilitating tasks such as summarizing voice recordings and suggesting replies in messaging apps.
As the AI race intensifies, Google's Gemini could be a potential game-changer, but it also faces fresh competition from Elon Musk's xAI and Chinese firm Baidu. Given the rapid evolution of AI technology, concerns about the potential risks of AI are also triggering global discussions on regulations and safe development.
ByteDance, the Tiktok owner, is launching new custom AI chatbot creation tools via a new soon-to-be-launched platform that will give users the ability to create their own custom AI bots that can respond to text queries in-stream. Custom LLMs and bots have become a new trend in the Chinese market, with ByteDance rival Baidu also launching a chatbot development system. Read more
Instagram is testing out another way to encourage interaction within Stories. Some users are now seeing a new “Hype” option, which enables you to post a comment to a Story that other users can then also see in-stream. At present, any replies to a Story are sent to the creator as a DM, but this will provide another interactive option. Read more
Meta is rolling out end-to-end encryption for calls and messages across its Facebook and Messenger platforms. Encrypted chats, first introduced as an optional feature in Messenger in 2016, will now be the standard for all users. Such encryption means that no one other than the sender and the recipient, not even Meta, can decipher people's messages. Read more
Threads has started rolling out topic tags (without the hash symbol) on the platform globally following a test in Australia last month. Unlike hashtags, people can use a phrase with space in it along with special characters. Users can click on a tag in a post to open all the posts in the search view. Users will only be able to tag one topic per post to avoid tag spam. Read more