Amid "dead internet" theories and AI bots across all social media, Meta wants to pivot to driving user engagement with AI users of its own - but is it what the internet really wants? Other key highlights of the week include:
- OpenAI makes its for-profit entity transition plans public in a blog post with stark opposition
- DeepSeek releases DeepSeek V3 to become China's most advanced and cost-effective open-source AI model
- Nvidia looks to shift part of its focus to robotics in 2025 as the industry reaches a "tipping point"
This is the last AI Tangle release for this year - thank you for sticking with us, and here's to an even grander 2025! ✨
|
AI may have been all the rage in 2024, but Facebook and Instagram parent company Meta thinks it should be taken a step further in 2025 with a controversial idea - AI bots on social media. Though traditional bots and ones leveraging AI recently on social media aren't anything new, they have always been created by third parties or malicious actors, never by the parent company themselves. Meta seems to want to be one of the first to jump aboard that train, but is it one we need?
What's Meta's aim here?
According to Connor Hayes, Meta's vice president of product for generative AI, these AI bots will exist as accounts with filled-out bios and all, aiming to increase user engagement, something they have been using AI to pivot towards since at least July. Though Meta reports that, across Facebook, Instagram, WhatsApp, and Messenger, at least one of its apps was used by 3.29 billion users daily, the company's organic growth has stagnated when looking at the 2.93 billion figure two years prior. Despite significant investments, other AI ventures, such as the Metaverse, have struggled to gain traction. Meta sees AI bots as a new way to sustain its relevance among younger audiences and create novel social media experiences, but the mere idea of a "dead internet" being endorsed by an industry leader is facing loud criticism across the web.
|
OpenAI's plans to transition to a for-profit is no big secret, but the move was recently made public in a blog post, with the goal of becoming a Delaware Public Benefit Corporation (PBC) to align its business model with its AGI mission. The PBC structure would balance shareholder, stakeholder, and public interests while enabling easier fundraising. The nonprofit arm would continue pursuing charitable initiatives, while the PBC would oversee operations. However, critics like Elon Musk and Meta don't plan to make the move easy, raising concerns by citing potential conflicts with OpenAI's original mission and governance transparency. Some former OpenAI employees, too, are questioning the prioritization of safety in the transition.
Late last week, Chinese startup DeepSeek unveiled its latest open-source large language model (LLM), DeepSeek V3, which, according to a report from the company, outperformed rival models like Meta's Llama 3.1 and Alibaba's Qwen 2.5 in benchmark tests on release. These results alone would make it best open-source model so far, additionally matching OpenAI’s GPT-4o and Anthropic's Claude 3.5 Sonnet. Trained on an incredibly modest budget of $5.58 million with just 2.78 million GPU hours on Nvidia H800 GPUs, DeepSeek V3 is a hint of China's prowess and how it is catching up with US-based AI tech.
With the upcoming launch of Jetson Thor, a compact computer for humanoid robots, Nvidia will be expanding into robotics in early 2025, as reported by the Financial Times. The company aims to lead in the "physical AI" market, which it sees at a "tipping point." While robotics contributes a smaller share to Nvidia's revenue compared to data centers, it is a strategic focus as competition in AI chips grows. Meanwhile, MIT's PRoC3S technology advances warehouse robotics, which allows safer handling of complex tasks previously requiring human dexterity.
According to an article by Bloomberg, Ibiden Co., a key supplier of chip package substrates for Nvidia, AMD, Samsung, and especially Intel, faces an ever-increasing demand for its AI-use substrates. As such, the company is expanding capacity by opening a new factory in Japan but may accelerate its plans to meet customer needs. Due to the difficult nature of mass-producing sophisticated substrates required for chips such as Nvidia's AI ones, Ibiden remains in a dominant position in the market. Even though its most valued partner Intel has been struggling recently, Ibiden CEO Koji Kawashima says the company will stay committed to it.
Brazilian startup Carecode, backed by a $4.3 million pre-seed round led by a16z and QED, aims to dampen healthcare costs and improve results by bringing in AI agents. Co-founded by Thomaz Srougi, founder of Dr. Consulta, and Pedro Magalhães, Carecode focuses on streamlining processes surrounding medical appointments. The AI system integrates with WhatsApp to accommodate Brazil's demographics, helping healthcare companies reduce call center costs and fill canceled slots proactively. Carecode targets a $100 billion market in Brazil, with plans for future expansion into related industries like insurance and payments.
|
Transistor - Get the most out of your favorite podcasts with Transistor by automatically converting your audio into engaging, searchable, and accessible text.
Hexus - Create interactive product demos, videos, step-by-step guides, and more in minutes with Hexus, an easy-to-learn stack for driving engagement and conversion.
Redactable - Ideal for handling confidential documents between businesses and agencies alike, Redactable is a cloud-based AI tool that helps organizations securely and efficiently remove private information from PDF documents.
Echo - Think deeper with Echo, an AI-powered voice and text note-taking app that refines your thoughts and brings your notes to life with smart organization.
|
AI Could Destabilize The US Power Grid in More Ways Than One (2-min read)
AI data centers could destabilize the US power grid, as Bloomberg reports a link between proximity to these centers and "bad harmonics," or disrupted electricity flow. Such distortions may harm appliances, increase fire risks, and cause outages, with AI centers' volatile energy demands exacerbating the problem.
|
| What did you think of this newsletter? Let us know! |
|
|
|
|
|