On December 6, Google introduced its latest breakthrough in artificial intelligence called Project Gemini, aiming to create an AI model that mimics human behavior. This move is expected to stir discussions about the potential benefits and risks of AI technology.
The launch will happen in stages, starting with simpler versions of Gemini named “Nano” and “Pro.” These will be integrated into Google’s AI-powered chatbot Bard and the Pixel 8 Pro smartphone.
With Gemini’s assistance, Google envisions Bard becoming more intuitive and proficient at tasks involving planning. On the Pixel 8 Pro, Gemini will quickly summarize recorded content on the device and offer automatic replies on messaging platforms, starting with WhatsApp. This innovation aims to enhance user experience and make technology more user-friendly.
In early 2024, Google plans to unveil “Bard Advanced,” an enhanced version of its chatbot, powered by the Ultra model of Project Gemini. However, this upgraded chatbot will be initially limited to a select test audience.
During its debut, Bard Advanced will operate exclusively in English worldwide. Nonetheless, Google executives have reassured reporters in a briefing that the technology is expected to smoothly expand its capabilities to encompass other languages in the future. This development highlights Google’s commitment to refining and diversifying its AI technology for a broader global audience.
Google demonstrated Gemini for a group of reporters, showcasing the potential of “Bard Advanced” in AI multitasking. This advanced chatbot can simultaneously recognize and understand presentations involving text, photos, and video, setting it apart in terms of capability.
Gemini is set to be integrated into Google’s search engine, although the exact timing for this transition hasn’t been specified yet.
Demis Hassabis, the CEO of Google DeepMind, the AI division responsible for Gemini, declared, “This marks a significant milestone in AI development and the beginning of a new era for us at Google.” Almost a decade ago, Google outbid other contenders, including Meta (Facebook’s parent company), to acquire London-based DeepMind. Since then, they’ve merged it with their “Brain” division to focus on advancing Gemini.
What is the debate?
Google is highlighting the problem-solving prowess of its technology, particularly its proficiency in math and physics. This has sparked optimism among AI enthusiasts who believe that such capabilities could pave the way for scientific breakthroughs that enhance human life.
However, on the flip side of the AI debate, concerns are growing about the potential for this technology to surpass human intelligence. This raises fears of significant job losses and the possibility of more harmful consequences, such as the amplification of misinformation or the inadvertent triggering of nuclear weapons. The debate around AI revolves not just around its potential benefits but also the risks and ethical considerations associated with its rapid advancement.
“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 introduction is poised to intensify the ongoing competition in the rapidly escalating field of AI. Over the past year, this competition has been heating up, involving key players such as the San Francisco startup OpenAI and the longstanding industry rival Microsoft. As each participant strives to advance their AI capabilities, the emergence of Gemini adds a new dimension to this dynamic landscape, signaling a potential shift in the balance of power within the AI industry.
Gemini to take on OpenAI’s GPT-4
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 2023 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 startups 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 startup.
So far, the partnership has proven lucrative for Microsoft, witnessing a remarkable surge of over 50% in market value in 2023. This growth is largely attributed to investors’ optimism that AI will become a lucrative frontier for the tech industry. Alphabet, Google’s parent company, has experienced a similar upward trend, with its market value soaring over $500 billion, approximately 45% higher this year.
Despite the high expectations surrounding Gemini, Alphabet’s stock saw a slight dip in trading on December 6. Microsoft’s increased collaboration with OpenAI over the past year, combined with OpenAI’s more assertive moves to commercialize its products, has led to concerns that the nonprofit organization may be veering from its original mission of safeguarding humanity as technology advances.
These concerns escalated in November 2023 when OpenAI’s board suddenly ousted CEO Sam Altman amid trust-related disputes. The backlash prompted fears of company destabilization and a potential mass departure of AI engineering talent to Microsoft. In response, OpenAI reinstated Altman as CEO and underwent a board reshuffle to address these concerns.
The emergence of Gemini puts OpenAI in a position where it must demonstrate that its technology remains superior to Google’s. Eli Collins, the vice president of product at Google DeepMind, expressed admiration for Gemini, stating, “I am in awe of what it’s capable of.”
During a virtual press conference, Google chose not to disclose Gemini’s parameter count, a metric used to measure a model’s complexity. However, a white paper released on December 6 highlighted the superior performance of the most capable version of Gemini compared to GPT-4 in multiple-choice exams, grade-school math, and other benchmarks. Nevertheless, the paper also acknowledged ongoing challenges in achieving higher-level reasoning skills with AI models.
Some computer scientists caution against overreliance on large language models, like those used in Gemini, which predict the next word in a sentence and are susceptible to errors known as hallucinations. Collins acknowledged progress in factuality with Gemini but noted that it remains an unsolved research problem, highlighting the continuous challenges in advancing AI models to achieve higher-order reasoning skills.