• NATURAL 20
  • Posts
  • Musk's Aim for Greater Control in Tesla's AI Future

Musk's Aim for Greater Control in Tesla's AI Future

PLUS: Vodafone & Microsoft 10-Year Deal, Adobe's AI-Powered Audio Editing and more.

Today:

Musk Pressures Tesla’s Board for Another Massive Stock Award

Elon Musk, CEO of Tesla, aims to increase his voting control in the company to 25% to feel comfortable leading its growth in artificial intelligence and robotics. He currently owns about 13% of Tesla's stock and has warned that without more influence, he might develop AI and robotics outside Tesla.

 Musk's focus includes Tesla's "Full Self-Driving" software and its prototype humanoid robots, though Tesla primarily earns from its automotive business. This move has sparked discussions about governance and his role as CEO, especially considering his recent Twitter acquisition and comments on share structure and compensation​.

Vodafone and Microsoft sign 10-year strategic partnership to bring generative AI, digital services and the cloud to more than 300 million businesses and consumers

Vodafone and Microsoft just teamed up for a massive 10-year deal. They're bringing cutting-edge AI, digital services, and cloud tech to over 300 million folks, mostly in Europe and Africa. This partnership is big news - it's gonna revolutionize how customers experience Vodafone services using Microsoft's super smart AI. 

They're also ramping up Vodafone's new IoT (Internet of Things) gig, making it a separate business. Big focus on M-Pesa too, aiming to make financial stuff more accessible in Africa. For small and medium businesses, they're rolling out new Microsoft tools to help them grow. And there's a massive shift for Vodafone, moving their data centers to Microsoft Azure, which should make things smoother and more efficient. 

Vodafone's dropping $1.5 billion over 10 years on this, and Microsoft's using Vodafone's connectivity services. They've got a bunch of cool projects lined up, like making customer interactions smoother with AI, expanding IoT services, boosting M-Pesa in Africa, helping small businesses with Microsoft's cloud services, and overhauling Vodafone's data centers with Azure.

Square Enix says it used AI art in upcoming Foamstars game

Square Enix's new game, Foamstars, is making waves by using AI-made art. Kosuke Okatani, the producer, shared that they used an AI tool, Midjourney, for some in-game album covers. But, he assured that the game's key parts, like gameplay, are all handmade. The company's boss, Takashi Kiryu, had earlier said they're big on using AI in game-making. 

This move's part of a bigger trend where AI's getting mixed into gaming, like Xbox creating AI characters. But, it's not all smooth sailing. There's legal heat about AI tools using artists' work without permission, and worries about AI replacing human jobs. In Foamstars, though, AI's role was just for some cool abstract images for album covers.

Adobe’s new AI-powered Premiere Pro features eradicate boring audio editing tasks

Adobe's spicing up Premiere Pro with some slick new AI tools that zap the hassle from audio editing. Say goodbye to the grind of hunting down tools or fixing crummy dialogue. The beta's rolling out interactive fade handles, letting you craft smooth audio transitions right on the main timeline. Plus, the Enhanced Speech feature, cleaning up junky dialogue, is hitting all users soon.

They're also dropping a neat trick for tagging audio clips. Whether it's dialogue, tunes, sound effects, or background noise, it gets tagged and easy to find. This cuts down on the click-fest, making editing smoother.

Waveforms, those squiggly sound lines, now adjust with track height, and clips have fresh colors for better visibility. All this jazz is in the Premiere Pro beta, open to anyone with Adobe Creative Cloud. It sits alongside the main app, so you can play with new features without messing up your usual setup.

🧠RESEARCH

The paper "Object-Centric Diffusion for Efficient Video Editing" introduces a method called Object-Centric Diffusion (OCD) for video editing. This method focuses on enhancing the editing process by allocating more resources to the foreground, which is crucial for perceptual quality. It proposes two novel techniques: Object-Centric Sampling and Object-Centric 3D Token Merging. These techniques reduce memory and computational costs significantly, enabling up to 10 times faster video editing without compromising quality. This approach is particularly effective in applications involving text-driven video edits. 

The paper "Diffusion Priors for Dynamic View Synthesis from Monocular Videos" presents a method to improve dynamic novel view synthesis. It addresses challenges in distinguishing motion and structure in videos, especially when camera poses are limited. The method involves finetuning a pretrained RGB-D diffusion model on video frames, then distilling the knowledge to 4D representations that combine dynamic and static Neural Radiance Fields (NeRF). This approach ensures geometric consistency and scene identity preservation, showing promising results in complex cases. 

The paper "TOFU: A Task of Fictitious Unlearning for LLMs" tackles the issue of unlearning in Large Language Models (LLMs). It introduces TOFU, a benchmark for assessing how well models can forget information. This is crucial to protect private data in LLMs trained on web data. TOFU provides a dataset with synthetic author profiles for testing unlearning, and it evaluates several existing unlearning methods. The results show that current methods are insufficient, highlighting the need for more effective unlearning techniques. 

🛠️TOP TOOLS

Now&Me - offering therapy and mental health support. It connects users with industry experts for various concerns like anxiety, stress, and sadness, providing 24x7 online support in multiple languages. 

CrystalSound - AI-powered app that offers real-time noise cancellation and voice-changing capabilities for Windows, Mac, and Linux. 

MindPal - AI-powered platform designed to enhance productivity and knowledge management. It allows users to build and utilize AI agents for automating tasks, integrating various data types like PDFs, Word, Excel, audio, video, and more. 

SkimAI - SkimAI is an AI-driven productivity tool designed to transform email management and enhance efficiency. It features AI-driven scheduling, custom workflows, and automatic email drafting, simplifying communication and decision-making processes. 

Dinnerfy -  simplifies meal planning and grocery shopping. It automates dinner planning, transforming user preferences into a sorted, editable, and shareable grocery list, and even offers 1-click Instacart delivery. 

📲SOCIAL MEDIA

🗞️MORE NEWS

Democratic inputs to AI grant program: lessons learned and implementation plans

OpenAI's latest initiative, the Democratic Inputs to AI Grant Program, turned heads by funding 10 global teams to craft AI governance tools with public input. They sifted through almost 1,000 proposals, picking diverse teams skilled in various fields. These teams tackled digital inclusivity and diverse representation, creating prototypes like AI-augmented deliberation platforms and systems for crowdsourcing AI model audits. Key takeaways? Public opinion is fickle, bridging the digital divide is tough, and finding middle ground in polarized groups is challenging. OpenAI plans to build on this, integrating these insights into their AI models. They're now on the lookout for bright minds to join this journey. OPENAI

The AI phones are coming

Samsung's upcoming Galaxy Unpacked event is generating buzz, especially for their promise of the most AI-centric phones yet. This follows a trend set by CES 2024's focus on integrating ChatGPT into various devices. While AI has been transformative in many sectors, its application in mobile phones has been limited so far. Current tech constraints mean that powerful language models like ChatGPT can't operate directly on phones and require cloud support. THE VERGE

AI can mimic handwriting because what else hasn’t it tried yet?

AI's capability to mimic handwriting marks another leap in its ever-expanding skill set. Researchers at Mohamed bin Zayed University of Artificial Intelligence in Abu Dhabi have created an AI model that learns to imitate handwriting by analyzing just a few paragraphs of handwritten text. This technology holds promise for aiding individuals who are unable to write manually, offering them a new way to express themselves in their own handwriting style. However, the researchers also recognize the potential misuse of this technology in forgeries, a concern that needs to be addressed as the technology develops. Meta previously introduced TextStyleBrush in 2021. TextStyleBrush is a text replacement model capable of altering words in images while retaining the original style, showcasing the increasing sophistication of AI in understanding and replicating complex human traits like handwriting. THE VERGE

How Doctors Are Using AI To Diagnose A Hidden Heart Condition In Kids

Doctors are harnessing AI to revolutionize the diagnosis of rheumatic heart disease (RHD) in children, a condition predominantly affecting impoverished countries like Uganda. Researchers at Children’s National Hospital, collaborating with the Uganda Heart Institute, have developed an AI-powered tool to diagnose RHD early, a critical step considering the disease’s deadly progression. With this AI tool, trained nurses can screen children for RHD efficiently and inexpensively, using technology that’s both portable and affordable compared to standard hospital echocardiogram machines. This early diagnosis enables treatment with penicillin, potentially saving thousands from advanced-stage surgeries. THE WASHINGTON POST

Goodnotes Acquires Korean AI Startup, Dropthebit, to Accelerate Development of AI Features

Goodnotes, a leading digital paper app, has acquired Korean AI startup Dropthebit, known for its service Traw. Traw specializes in organizing meeting, lecture, and video content into summarized notes and videos. This acquisition will integrate Traw's generative AI technology into the Goodnotes platform, enhancing its features. Since 2020, Traw has been automating note creation from various sources. With initial funding from Capstone Partners and KB Investment, Traw launched a smart digital whiteboard in 2022, capable of recording and summarizing meetings and lectures. In May, Traw AI was introduced to summarize YouTube content effectively. PR NEWSWIRE

Delfos Raises $6.9 Million To Streamline Clean Energy With AI

Delfos, an AI energy software startup, has successfully raised €6.3 million ($6.9 million) in a seed funding round to expand its operations across Europe. Based in Barcelona, the company has developed an innovative AI and big data-based virtual engineer for the renewable energy sector. This technology automates performance and workflow while ensuring reliability in renewable energy operations. The funding round saw participation from climate tech-focused venture capital firm Contrarian Ventures, Headline, DOMO.VC, and EDP Ventures. Delfos' software currently monitors over 10GW of renewable energy. The company claims its technology can enhance operational efficiency and reduce the cost of component replacement by 30%, while also boosting energy generation by 3-5%. FORBES

NASA accelerates science with gen AI-powered search

NASA is leveraging generative AI and neural networks to tackle the challenge of navigating its massive data repositories. With an extensive network that includes seven operating centers, nine research facilities, and over 18,000 staff, NASA generates a vast amount of data across various scientific fields such as astrophysics, heliophysics, biological science, physical science, earth science, and planetary science. The agency currently stores more than 88,000 datasets and 715,000 documents across 128 data sources, with its earth science data alone expected to reach 250 petabytes by 2025. The move towards using a combination of neural nets and generative AI is part of NASA's effort to make these extensive data resources readily accessible to scientists and researchers. CIO

Researchers combine automated experiments with AI to boost drug development

Researchers at the University of College of Cambridge have developed a groundbreaking platform that integrates automated experiments with artificial intelligence (AI) to enhance drug development. This platform is designed to provide deeper insights into the interactions of various chemicals, which is a critical aspect of discovering and producing new medicines. Cambridge team has adopted a data-driven approach, inspired by advancements in genomics. Their innovative system, referred to as the chemical 'reactome,' employs a combination of automated experiments and machine learning. This approach significantly accelerates the process of understanding chemical reactivity. INTERESTING ENGINEERING

Microsoft’s Nadella Wants Stability at OpenAI, Not Control

Microsoft CEO Satya Nadella has expressed contentment with the current relationship and structure between Microsoft and OpenAI, emphasizing a desire for stability rather than increased control. This statement comes amid potential scrutiny from UK and EU watchdogs who are considering investigating the close ties between the two companies. During an interview at the World Economic Forum in Davos, Nadella addressed the partnership with OpenAI, indicating that Microsoft's investment and partnership were established regardless of OpenAI's nature as a capped-profit or nonprofit organization. Nadella's remarks reflect a position of satisfaction with the status quo, underscoring his confidence in the partnership's current setup without seeking further control over OpenAI. BLOOMBERG

Bill Gates explains how AI will change our lives in 5 years

Bill Gates, in an interview with CNN's Fareed Zakaria, expressed his optimism about the transformative impact of artificial intelligence (AI) on society within the next five years. He acknowledged the widespread concern about AI potentially affecting 40% of global jobs, as reported by the International Monetary Fund, but emphasized the historical pattern of new technology initially causing fear, followed by the creation of new opportunities and jobs. Drawing a parallel with the agricultural revolution of the 1900s, Gates suggested that AI, like past technological advancements, will lead to the emergence of new job categories and overall societal improvement. He particularly highlighted AI's potential to ease the burdens of administrative tasks in professions like medicine, where it could efficiently handle doctors' paperwork. CNN

OpenAI CEO Altman says at Davos future AI depends on energy breakthrough

Sam Altman, CEO of OpenAI, spoke at the World Economic Forum's annual meeting in Davos, emphasizing the need for an energy breakthrough for the future of AI. He mentioned that AI will require much more power than anticipated, and pointed to climate-friendly energy sources like nuclear fusion or cheaper solar power and storage as key solutions. Altman, who personally invested in U.S. nuclear fusion company Helion Energy, believes that such an energy breakthrough is essential for AI's advancement. He also noted his wish for the world to embrace nuclear fission as an energy source. Microsoft, OpenAI's biggest financial backer, has a deal with Helion Energy for future energy supply​. REUTERS

Windows 11 Paint has its own AI image generator tool - here's how to use it

Microsoft's Paint in Windows 11 has a new AI feature called Cocreator. You describe what you want, pick a style (like watercolor or photorealistic), and Cocreator generates three images.

How to Use:

  • Update Windows 11 to the latest version.

  • Update Microsoft Paint through the Microsoft Store.

  • Open Paint, choose a blank canvas, and click the Cocreator button.

Creating Images:

  • Describe what you want in the Cocreator pane.

  • Pick a style like Charcoal, Watercolor, Oil Painting, etc.

  • Hit Create to generate three images based on your description and style.

Editing and Saving:

  • Zoom and view the generated images.

  • Edit using Paint's tools if needed.

  • Save in formats like PNG, JPG, etc., or send via email.

This is a neat tool for whipping up images in a snap, giving you control over style and subject without the worry of copyright issues.

    What'd you think of today's edition?

    Login or Subscribe to participate in polls.

    What are MOST interested in learning about AI?

    What stories or resources will be most interesting for you to hear about?

    Login or Subscribe to participate in polls.

    Join the conversation

    or to participate.