Google Launches Project Gemini: A Significant Advancement in AI
With the introduction of project Gemini, an AI model taught to behave in human-like ways, Google made its next major advancement in artificial intelligence yesterday. This move is expected to heighten discussion regarding the benefits and drawbacks of the technology.
Phases of the distribution will take place: Google’s AI-powered chatbot Bard and its Pixel 8 Pro smartphone will launch with the less advanced Gemini versions, dubbed “Nano” and “Pro.”
Google claims Bard will grow more intuitive and proficient at planning tasks with Gemini’s assistance. Google claims that Gemini on the Pixel 8 Pro will be able to automatically respond to messages on messaging apps, like as WhatsApp, and swiftly summarize recordings made on the device.
The largest developments from Gemini won’t be seen until early in the upcoming year, when the company’s Ultra model will be utilized to introduce “Bard Advanced,” a more advanced chatbot that will initially only be made available to a test group.
Although Google executives reassured reporters during a briefing that the technology will have no trouble gradually spreading into other languages, the AI will initially only function in English globally.
A demonstration of Gemini for a group of reporters suggested that Google’s “Bard Advanced” could be able to recognize and comprehend presentations including text, images, and video at the same time—an unparalleled level of AI multitasking.
Additionally, Gemini will soon be integrated into Google’s widely utilized search engine, though the exact date of that change has not yet been announced.
Google DeepMind CEO Demis Hassabis said, “This is a significant milestone in the development of AI, and the start of a new era for us at Google.” Google DeepMind is the company that created Gemini.
Nearly ten years ago, Google beat other rivals, including Facebook parent Meta, to acquire London-based DeepMind. Since then, it has combined DeepMind with its “Brain” division to concentrate on the advancement of Gemini.
Google has been promoting the technology’s problem-solving abilities as being particularly strong in arithmetic and physics, which has raised hopes among proponents of AI that it would result in scientific discoveries that will enhance humankind’s quality of life.
However, some AI skeptics fear that eventually AI will surpass human intellect, leading to the loss of millions of jobs and possibly even more destructive behavior like spreading false information or inciting the use of nuclear weapons.
“We’re approaching this work boldly and responsibly,” Google CEO Sundar Pichai wrote in a blog post. “That means being ambitious in our research and pursuing the capabilities that will bring enormous benefits to people and society, while building in safeguards and working collaboratively with governments and experts to address risks as AI becomes more capable.”
With the launch of Gemini, the AI race between Microsoft and San Francisco company OpenAI, which has been intensifying over the past year, is expected to heat up.
When OpenAI published the free ChatGPT tool late last year, it was already well into constructing its most advanced AI model, GPT-4, with the financial and computational support of Microsoft. As a result of the AI-powered chatbot’s meteoric rise to stardom, generative AI’s business potential gained attention, and Google was forced to remove Bard.
OpenAI launched GPT-4 in March, just as Bard was making his appearance on the stage. Since then, the company has been adding new features targeted at consumers and corporate clients, such as the ability for the chatbot to analyze photos, which was introduced in November. It has been vying for business with rival AI startups like Anthropic and even Microsoft, which has granted OpenAI the exclusive rights to use its technology in return for Microsoft investing billions of dollars in the company.
Microsoft has benefited greatly from the collaboration thus far, with its market value rising by over 50% this year, mostly due to investors’ expectations that artificial intelligence will become a tech industry gold mine.
Alphabet, the parent company of Google, has also been riding high on this wave, with its market value jumping by more than $500 billion, or around 45%, so far this year.
Even though Gemini’s launch had been anticipated for several months, Alphabet’s shares did not move much during yesterday morning’s trade.
Concerns have been expressed about whether OpenAI has deviated from its original aim to preserve mankind as technology advances in light of Microsoft’s increasing engagement in the non-profit organization over the past year and OpenAI’s more aggressive attempts to commercialize its products.
These concerns were heightened last month when Sam Altman, the CEO of OpenAI, was unceremoniously ousted by the board over unreported trust difficulties. Following criticism that threatened to bankrupt the business and cause a large-scale departure of skilled AI engineers to Microsoft, OpenAI reinstated Altman as CEO and reorganized its board.
With the release of Gemini, OpenAI might find itself attempting to demonstrate that its technology is still superior than Google’s.
“I am in awe of what it’s capable of,” Google DeepMind vice president of product Eli Collins said of Gemini.
Google declined to provide Gemini’s parameter count, which is one indicator of a model’s complexity but not the only one, during a virtual press conference.
The most advanced version of Gemini, according to a white paper published yesterday, outperforms GPT-4 on multiple-choice tests, elementary school math, and other benchmarks. However, the research also highlighted that there are still issues with getting AI models to develop higher-order reasoning abilities.
Large language models, which operate by continually guessing the word that will come next in a sentence and are prone to making up mistakes known as hallucinations, are seen as having limitations by some computer scientists.
“We made a ton of progress in what’s called factuality with Gemini. So Gemini is our best model in that regard. But it’s still, I would say, an unsolved research problem,” Collins said.
