- Treky
- Posts
- AI Race Intensifies
AI Race Intensifies
META Joins AI Race
The current surge in generative AI stems from the belief in the tech industry that we are progressing towards creating superhuman intelligence.
OpenAI and Google's AI leader, Demis Hassabis, are both striving to develop Artificial General Intelligence (AGI). Mark Zuckerberg of Meta is also joining this race. He doesn't have a specific timeline or definition for AGI but aims to integrate Meta's AI research group, FAIR, with the team developing generative AI products. Zuckerberg believes that recruiting top AI talent is crucial for building products that require general intelligence.
Meta is heavily investing in computing power, with plans to possess over 340,000 Nvidia H100 GPUs by the end of the year, showcasing its substantial capacity in AI development. However, the exact nature and timeline of achieving AGI remain unclear. Zuckerberg views AGI's development as a broad spectrum of capabilities, including reasoning and intuition, and sees its arrival as a gradual process.
Meta's focus on AGI was partly influenced by the release of its Llama 2 language model. Despite Llama 2 not being industry-leading, it guided Meta's approach to building smarter AI, including coding capabilities and advanced reasoning. Zuckerberg aims to make Llama 3 and future models state-of-the-art.
The control over AGI is a contentious issue. Zuckerberg, with significant voting control over Meta, advocates for an open-source approach to AI development, contrasting with companies that have become less transparent. He sees the open-source model as a way to address inequalities in access to AI technology.
Meta's focus on AGI coincides with its commitment to the metaverse. Zuckerberg denies that this focus represents a pivot from the metaverse, emphasizing Meta's ongoing investment in virtual and augmented reality. He envisions a future where AI-generated virtual worlds and AI characters play a significant role in Meta's platforms.
Introducing Galaxy AI
Samsung recently launched the Galaxy S24 series, heavily emphasizing AI-powered features at its Galaxy Unpacked event. The new Galaxy S24 range, including the S24, S24+, and S24 Ultra, introduces various AI capabilities like circle to search, generative edit, and live translation for phone calls, mostly powered by Galaxy AI.
However, Samsung disclosed that these Galaxy AI features will only be free until 2025, after which they may require a subscription. The exact details of the subscription model are unclear, but it's noted that most Galaxy AI features operate on the cloud, except for the on-device live translation feature.
To promote the Galaxy S24 series, Samsung has set up Galaxy Studio pop-up stores across South Korea, where consumers can experience the AI features of the new devices firsthand. These studios, located in places like Samsung Gangnam and Samsung Store Hongdae, will operate for about a month.
They offer demonstrations of features like live language translation and AI-powered apps such as Samsung Notes and Voice Recorder. Special areas within these studios allow visitors to try out the Circle to Search feature and experience the AI-enhanced camera quality, especially in low-light conditions, through the ProVisual Engine.
Additionally, Samsung plans to collaborate with various brands and themes like fashion, food, and travel to showcase the full range of the Galaxy S24's AI functions.
UAE: A Potential AI Superpower
The UAE is emerging as a potential AI superpower, a result of two decades of strategic planning. The UAE's Technology Innovation Institute recently released a top-performing generative AI model, part of the Falcon series, signifying the country's advances in AI technology.
Omar Sultan Al Olama, the UAE's AI minister, outlines the country's ambitious plan to transform into a world leader in AI, leveraging its resources for comprehensive development in AI applications, ranging from government operations to core research and AI hardware manufacturing.
Al Olama's vision includes positioning the UAE as a leading technology provider, especially to the Global South, challenging China's "Digital Silk Road" and potentially diverging from American foreign policy. He emphasizes the importance of not missing out on technological revolutions. His strategy prioritizes quality of life improvements and infrastructure enhancement through AI, as well as promoting AI literacy.
The UAE's approach includes significant investments, such as controlling GlobalFoundries, one of the top semiconductor fabricators, and fostering AI education and leadership through partnerships with institutions like Oxford University. It has also established dedicated AI institutions, including a university and the Technology Innovation Institute.
Looking forward, the UAE plans to launch a state-backed AI company and integrate AI technology in upcoming space missions, including a moon mission in 2024. This comprehensive approach positions the UAE as a serious contender in the global AI landscape.
Introducing Microsoft Copilot
Microsoft has expanded the availability of its AI tool, Copilot, aiming to enhance document and email writing for a broader range of users.
Initially launched for large enterprises, Copilot is now accessible through a $20 monthly subscription called Copilot Pro, targeting consumers already using Microsoft's 365 suite.
Additionally, Microsoft introduced a new deal for small businesses, allowing them to buy up to 299 licenses at $30 per person per month, a shift from the prior minimum of 300 licenses. This move makes the service more approachable for small and medium-sized businesses.
Copilot, powered by OpenAI's large language models (LLMs), is integrated into Microsoft applications like Outlook, Word, Excel, and PowerPoint. Microsoft is not alone in this endeavor; competitors like Google with Bard for Workspace and Salesforce's Einstein Copilot are also integrating LLMs into workplace software. Microsoft's strategy, as explained by GM Divya Kumar, reflects feedback from smaller businesses and aims to address their needs, especially amidst staffing shortages. Kumar highlighted that one of the most appreciated features of Copilot is its ability to summarize complex text.
Despite its utility in business settings, Forrester Research analyst J.P. Gownder pointed out that pitching Copilot to individual consumers might be more challenging. Unlike businesses, consumers lack vast repositories of internal information, which Copilot leverages.
Gownder views Microsoft's rollout of Copilot as a significant step towards familiarizing more users with its generative AI technology in a work context, marking a crucial move in keeping up with competitors in the AI integration landscape.
New AI Device (R1)
Jesse Lyu, CEO of the AI startup Rabbit, introduces the R1, a $199 AI device aimed not to replace smartphones immediately, but with potential to do so in the future. The R1 sold out its initial 10,000 units and is now accepting pre-orders for a second batch.
The device resembles a Playdate console or a modern handheld TV, about half the size of an iPhone. It features a 2.88-inch touchscreen, rotating camera, scroll wheel/button, 2.3GHz MediaTek processor, 4GB memory, 128GB storage, and an all-day battery. Designed with Teenage Engineering, it's lightweight and user-friendly but prone to fingerprints.
R1's core is its Rabbit OS and underlying AI, a "Large Action Model" (LAM) serving as a universal app controller. It simplifies tasks like music control, car ordering, grocery shopping, and messaging through a single interface, without juggling apps and logins. Instead of developing new APIs, Rabbit trained LAM using existing apps, teaching it interface navigation and task execution.
The R1 features a training mode for teaching new tasks, promising to replicate them independently. It also offers a web portal, Rabbit Hole, for service integration and a virtual machine option for teaching the device.
While the R1 is not yet a smartphone replacement, it functions as a voice assistant with additional features like a screen and camera. It aims to be a multifunctional device, considering security and privacy in its design. Pre-orders are open, with shipments expected to start in March, potentially preceding Humane's AI Pin in the market.
ElevenLabs Raises $80M
ElevenLabs, a startup specializing in AI-powered synthetic voice technology, recently secured $80 million in a Series B funding round, led by notable investors such as Andreessen Horowitz and former GitHub CEO Nat Friedman. This funding boosts the company's total capital to $101 million and its valuation to over $1 billion. The funds will be used for product development, team expansion, AI research, and enhancing safety measures.
Co-founded by Piotr Dabkowski and CEO Mati Staniszewski in 2022, ElevenLabs aims to revolutionize voice cloning, inspired by the duo's dissatisfaction with poorly dubbed American films in Poland. The company offers a browser-based speech generation application, allowing users to create lifelike voices with customizable features. Their technology caters to diverse applications like audiobooks, film dubbing, game characters, and marketing activations.
One of ElevenLabs' notable achievements is a "speech to speech" tool designed for movies and TV shows, which maintains the original speaker's voice characteristics while removing background noise and synchronizing with translated speech. The company is also working on a dubbing studio workflow and a mobile app for narrating web pages and texts.
Despite these innovations, ElevenLabs has encountered controversies. Its technology was misused on platforms like 4chan for creating hateful messages and even for cloning voices convincingly enough to deceive a bank's authentication system. In response, ElevenLabs is enhancing its detection tools to flag abusive usage and plans to collaborate with third-party platforms to regulate the misuse of its technology.
Voice actors have also raised concerns, accusing ElevenLabs of using their voice samples without consent and fearing AI-generated voices might replace them. In contrast, Replica Studios, a competitor, has sought to balance AI advancements with voice actors' rights through an agreement with SAG-AFTRA.
To address these issues, ElevenLabs is developing a voice marketplace where creators can share their voices and receive compensation. However, this compensation is currently in the form of credits towards ElevenLabs' services, not cash.
ElevenLabs continues to grow and innovate in the synthetic voice market, competing with other startups and major tech firms while navigating the complex ethical and industry challenges in the field.