Google launches Gemini, upping the stakes in the global AI race

Gemini’s biggest advances will not come until early next year when its Ultra model will be used to launch “Bard Advanced”, a juiced-up version of the chatbot that initially will only be offered to a test audience.

Elon Musk’s xAI start-up seeks to raise US$1 billion in equity

The AI, at first, will only work in English throughout the world, although Google executives assured reporters during a briefing that the technology will have no problem eventually diversifying into other languages.

Based on a demonstration of Gemini for a group of reporters, Google’s “Bard Advanced” might be capable of unprecedented AI multitasking by simultaneously recognising and understanding presentations involving text, photos and video.

Gemini will also eventually be infused into Google’s dominant search engine, although the timing of that transition has not been spelled out yet.

“This is a significant milestone in the development of AI, and the start of a new era for us at Google,” declared Demis Hassabis, CEO of Google DeepMind, the AI division behind Gemini. Google prevailed over other bidders, including Facebook parent Meta Platforms, to acquire London-based DeepMind nearly a decade ago, and has since melded it with its “Brain” division to focus on Gemini’s development.

The technology’s problem-solving skills are being touted by Google as being especially adept in maths and physics, fuelling hopes among AI optimists that it may lead to scientific breakthroughs that improve life for humans.

OpenAI applies for GPT-6, GPT-7 trademarks in China

But an opposing side of the AI debate worries about the technology eventually eclipsing human intelligence, resulting in the loss of millions of jobs and perhaps even more destructive behaviour, such as amplifying misinformation or triggering the deployment 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.”

Gemini’s arrival is likely to up the ante in an AI competition that has been escalating for the past year, with San Francisco start-up OpenAI and long-time industry rival Microsoft.

Google’s DeepMind builds AI for more accurate weather forecasts

Backed by Microsoft’s financial muscle and computing power, OpenAI was already deep into developing its most advanced AI model, GPT-4, when it released the free ChatGPT tool late last year. That AI-fuelled chatbot rocketed to global fame, bringing buzz to the commercial promise of generative AI and pressuring Google to push out Bard in response.

Just as Bard was arriving on the scene, OpenAI released GPT-4 in March and has since been building in new capabilities aimed at consumers and business customers, including a feature unveiled in November that enables the chatbot to analyse images. It has been competing for business against other rival AI start-ups such as Anthropic and even its partner, Microsoft, which has exclusive rights to OpenAI’s technology in exchange for the billions of dollars that it has poured into the start-up.

Microsoft’s deepening involvement in OpenAI during the past year, coupled with OpenAI’s more aggressive attempts to commercialise its products, has raised concerns that the non-profit has strayed from its original mission to protect humanity as the technology progresses.

With Gemini coming out, OpenAI may find itself trying to prove its technology remains smarter than Google’s.

“I am in awe of what it’s capable of,” Google DeepMind vice-president of product Eli Collins said of Gemini.

In a virtual press conference, Google declined to share Gemini’s parameter count – one but not the only measure of a model’s complexity. A white paper released on Wednesday outlined the most capable version of Gemini outperforming GPT-4 on multiple-choice exams, grade-school maths and other benchmarks, but acknowledged ongoing struggles in getting AI models to achieve higher-level reasoning skills.

Some computer scientists see limits in how much can be done with large language models, which work by repeatedly predicting the next word in a sentence and are prone to making up errors known as hallucinations.

“We made a ton of progress in what’s called factuality with Geminim” Collins said. “So Gemini is our best model in that regard. But it’s still, I would say, an unsolved research problem.”