A.I. topped the list of big tech buzzwords in 2023, for good or for bad, and remained front and center Wednesday when Google revealed its advanced Gemini A.I. system. This latest development, which compounds the power of generative content, marks another step in making this truly mainstream technology. The general public is increasingly used to chatting to A.I.s, thanks to ChatGPT, but what sets Gemini apart is its "multimodality," which applies the program's artificial intelligence smarts to text, code, image queries, and more all at once. Current A.I. users may use one system for text, another for images, and so on. The benefits of these tools to small businesses are clear, of course: If a small company's PR department is already snowed under, for example, speedy A.I. text or image generation could be a boon. Gemini condenses these A.I. capabilities in one app, simplifying user requirements.
The other thing that sets Gemini apart, of course, is that it comes from Google, a technology giant whose tools we typically all use every day at home and at work. Google's global heft could make Gemini hard to ignore, even for users who are comfortable with ChatGPT. When announcing the launch on X, Google and Alphabet CEO Sundar Pichai said it was the company's "most capable and general A.I. model yet."
Users trying Gemini right now will discover its limits, which are short of what it ultimately promises to do. Google's event yesterday was arguably a soft launch to put its new A.I. tools in users' hands, with more to come later.
The current version, Gemini Pro, is integrated into the company's Bard chatbot (there's also a "lighter" version called Nano aimed at smaller, more agile applications such as phone and tablet apps). Essentially, Gemini Pro in Bard is a lot like ChatGPT, but with a different underlying A.I. infrastructure. This means users can now leverage Gemini for a smorgasbord of business needs -- whether it's writing copy for press releases, coming up with pleasing phrases to improve financial slideshows, or even helping developers write code. For those already using ChatGPT to generate some content for users, it would be easy to use similar prompts with Gemini via Bard, and then choose which A.I.'s output is preferable.
Google plans to launch the full Gemini Ultra system next year as part of Bard Advanced, which is when the technology could become even more transformative for businesses. Teasing the Gemini Ultra unveiling, Pichai said on X it "exceeds current state-of-the-art results." In response to the launch news, Elon Musk, whose companies have longstanding connections to A.I. applications, replied "Impressive."
The product demonstration emphasized Gemini's seamless capabilities to generate images, diagrams, and other content as well as text. This fusion has gigantic implications for the business world. One demo stood out: YouTuber Mark Rober designed a giant paper airplane with the help of Bard, including taking a photo of it and then asking for a critique. Bard examined the image, pointed out design flaws, and suggested improvements. This sort of all-in-one effort has clear benefits for small businesses, where workers' expertise with a single A.I. system could boost efficiencies simply by not requiring familiarity with several different A.I. technologies.
Since A.I. competition remains fierce and fast, Meta, Facebook's parent company, also on Wednesday announced its own standalone generative A.I. system. Called Imagine and released through Meta AI, it followed a familiar model. Users employ natural language to ask for an image and the A.I. "dreams up" its response. The results can be startling, beautiful, weird, or even seamlessly realistic. Akin to Google's A.I. push with Gemini, Meta's entry into this field means A.I.-generated imagery will likely become even more pervasive.
Also chiming in was Apple, whose machine learning research team announced MLX, a new framework for developers to access its machine learning systems running on Apple chips. Machine learning is not quite "full A.I.," but focuses on using algorithms and statistics to analyze data patterns to build smart systems. The new framework means developers will be able to improve the way that they train A.I. libraries used in other apps so they run more smoothly on Apple devices. Apple has long been considered more cautious in developing A.I. than other companies, hence this focus on the back-end systems rather than customer-facing products.
The impact of three global tech giants announcing A.I. developments in a single day also combines to mean one thing: Business use cases for A.I. will explode. While these applications may well allow enterprises to cut costs, boost innovation, or smooth out business processes, A.I.s bring their own set of complex issues.
A.I. models are well known to be imperfect, as shown by previous reports of Meta's problems with racial bias in some A.I. images. There's also the issue of intent, as some A.I.-generated content is problematic, from celebrity deepfakes that have already spawned lawsuits, and confounding faked images of the war in Gaza complicating public perceptions of the conflict. There are also complex questions of A.I. transparency, including whether A.I.-generated content should be clearly labeled. Meta, for example, uses hidden watermark technology in its new Imagine images so that they are detectable later, leaving unknown the implications for businesses that use the images. A.I. has also been much discussed by government bodies around the world, with the EU now working on rules and regulations for the technology.
But with Gemini set to get still more clever, and corporate giants like McDonald's set to embrace A.I. in an effort to transform their businesses, 2024 looks to be the year the technology goes truly mainstream.
No comments:
Post a Comment