Musk's Cyborg Vision, VR's Fate, Quantum Computing's Impact, and More

The Morning Brief

In Today’s Edition:

Elon Musk envisions a cyborg body, VR's struggle, and Quantum Computing's impact on work. OpenAI's customized ChatGPT, Mastercard's Web3 music NFTs, and Meta's Quest 3 with generative AI on the horizon.

Stay informed with today's top metaverse tech stories and gain access to over 15 quick bytes, research papers, tools, guides, and resources to help you navigate the metaverse.

Must-reads

  1. Elon Musk Envisions Tesla and Neuralink Building a Cyborg Body for Amputees: Elon Musk has proposed a collaboration between his companies, Tesla and Neuralink, to design prosthetic limbs for amputees. He believes that by combining a Neuralink implant with a robotic arm or leg, they can create a cyborg body that is incredibly capable. However, both companies are still in the early stages of their respective projects, with Tesla only beginning to develop a humanoid robot in 2021 and Neuralink recently receiving FDA approval for a first human clinical study for its brain-machine interface chip. (Fortune)

  2. VR is Dead: Despite the initial hype, VR is struggling to find a compelling use case, leading to the assertion that VR is essentially dead. Even the cheapest mainstream headset, the Meta Quest 2, has AR capabilities, and Apple's Vision Pro has moved so far from the VR concept that it doesn't even use the VR nomenclature anymore. The VR market, which was predicted to hit $150 billion by 2020, sits at around $32 billion nearing 2024, a fifth of what was initially projected. (TechCrunch)

  3. Andreessen Horowitz Exec Sees Promising Paths for Web3 Social Platforms: Sriram Krishnan, a partner at Andreessen Horowitz, believes that Web3 has the solutions to the problems found on Web2-based social media platforms. He suggests that now is the time to build large-scale crypto consumer social media marketplace applications. (Cointelegraph)

  4. Quantum Computing's Impact on the Future of Work: Quantum computing, which uses principles of quantum theory to process vast quantities of information in moments, is leading the next technological revolution. It has potential applications in manufacturing, artificial intelligence, cybersecurity, financial institutions, pharmaceutical industry, engineering, aerospace, military development, and domestic power sources. However, businesses will need to adjust their preexisting security protocols, networks, and data processing platforms to keep up with the new demands. Despite challenges such as sensitivity to environmental factors and the need for human double-checking, venture capital investments in quantum computing grew by 500% from 2015 to 2020. Businesses are encouraged to invest in quantum computer infrastructures, conduct employee training, adopt new security protocols, and map out development and adoption strategies. (Fast Company)

  5. OpenAI's ChatGPT Moat in AI Wars Shrinking: OpenAI's ChatGPT has been a dominant force in the AI space since its launch, but the release of Meta's Llama 2 poses a significant threat. Llama 2, a large AI model, is freely available to developers for research and commercial use, which could potentially erode ChatGPT's competitive advantage. Additionally, there are growing concerns about the quality of responses from GPT-4, the advanced AI model underlying the premium version of ChatGPT. As the novelty of ChatGPT wears off, OpenAI will need to address these issues to maintain its position in the rapidly evolving AI landscape. (Business Insider)

  6. OpenAI Launches Customized Instructions for ChatGPT: OpenAI has introduced a new feature that allows users to set custom instructions for ChatGPT. This feature enables users to share any context they want ChatGPT to consider in its responses, eliminating the need to repeatedly set the same instructions. For instance, a teacher can specify they are teaching fourth-grade math, or a developer can specify their preferred coding language. This feature is currently in beta and available for Plus plan users, excluding those based in the EU and the U.K. The information provided for custom instructions will be used to train API models to adapt to different instructions. (TechCrunch)

  7. Mastercard's Web3 Music Accelerator Drops Five New Singles as NFTs: Mastercard's Web3 music program, The Mastercard Artist Accelerator, is set to culminate with a showcase featuring five artists from around the world. The live-streamed show will debut the AI-driven singles produced through the Artist Accelerator program, which was launched to equip emerging artists with the tools, skills, and access to forge their own musical paths in the digital economy. Each track was created using the AI Music Studio, powered by WarpSound, a leading generative AI music technologies company. Each song will be available to claim as an NFT on the OneOf Marketplace on July 27, during the showcase. (Entrepreneur)

  8. Leaders Warned Against Using ChatGPT for Data Analytics: Brandon Southern, former head of analytics at eBay, Amazon, and GameStop, warns that using ChatGPT for data analytics is a risky move. He argues that AI can't replace the work of data analysts due to the complexity and multiple sources of truth in data. (Business Insider)

  9. McDonald's Opens McNuggets Land in the Metaverse: McDonald's is celebrating the 40th anniversary of Chicken McNuggets by opening McNuggets Land in the metaverse platform, The Sandbox. The experience, spearheaded by the Hong Kong arm of McDonald's, invites players to find four McDonald’s signs to earn rewards including a shared prize pool and mystery boxes. Hong Kong users also have the chance to win coupons and a 365-day free Chicken McNuggets prize. Despite the gamified experience, the initiative has raised questions about its timing and relevance in the current Web3 environment. (CoinDesk)

  10. Meta Quest 3 Could Gain Generative AI Power in 2024: Meta's upcoming Quest 3 could be powered by generative AI, according to reports. The integration of generative AI could significantly enhance the capabilities of the Quest 3. (Mixed News)

🦄 Startups & Venture Capital

  1. Karat Raises $70M for Financial Tools for Content Creators: Karat Financial, a startup offering credit to creators, has raised $70 million in venture capital. The company provides a business credit card that offers creators higher limits based on their social and financial stats instead of FICO scores. The new funds will be used to scale its existing products and continue to build tools to help creators manage money. (TechCrunch)

  2. Polychain Capital and CoinFund Raise Crypto-Focused VC Funds: Polychain Capital and CoinFund have raised significant funds for their crypto-focused venture capital initiatives. The funds will be used to invest in promising blockchain and cryptocurrency startups. (Axios)

  3. Spiritt Raises Seed Funding to $13.5M for AI Platform that Builds Apps Through Verbal Descriptions: Spiritt, an Israeli startup, has extended its seed funding round to $13.5 million. The company's AI platform allows individuals with no programming knowledge to independently develop complex applications by verbally describing their ideas to the computer. The funds will be used to expand the team and support clients in their business growth, with a focus on the U.S. market. (CTech)

  4. Binance Labs Invests $10M in DeFi Lender Radiant: Binance Labs has invested $10 million in Radiant Capital, a decentralized finance (DeFi) lending and borrowing protocol. The funds will be used for tech and product development, including expanding collateral and deployment on the Ethereum mainnet. (CoinDesk)

Quick Bytes

Academic Papers and Research

  • GLSFormer : Gated - Long, Short Sequence Transformer for Step Recognition in Surgical Videos: In this paper, the authors propose a new approach called GLSFormer (Gated - Long, Short Sequence Transformer) for automated surgical step recognition in videos. Existing methods either separately model spatial and temporal information or operate on short-range temporal resolution. GLSFormer, however, uses a vision transformer-based approach to jointly learn spatio-temporal features directly from sequences of frame-level patches. The method incorporates a gated-temporal attention mechanism, which intelligently combines short-term and long-term spatio-temporal feature representations. The approach is extensively evaluated on two cataract surgery video datasets, and it outperforms various state-of-the-art methods. The results demonstrate the effectiveness of GLSFormer for automated surgical step recognition, which can significantly enhance patient safety and decision-making during surgeries. Link to code

  • AlignDet: Aligning Pre-training and Fine-tuning in Object Detection: The common practice of large-scale pre-training followed by fine-tuning in object detection algorithms can lead to discrepancies in data, model, and task, limiting the detector's performance, generalization ability, and convergence speed. To address this, the authors propose AlignDet, a unified pre-training framework that decouples pre-training into two stages: image-domain pre-training and box-domain pre-training. The former optimizes the detection backbone for holistic visual abstraction, while the latter learns instance-level semantics and task-aware concepts to initialize parts outside the backbone. AlignDet incorporates self-supervised pre-trained backbones, enabling unsupervised pre-training of all modules for various detectors. Extensive experiments demonstrate significant improvements in various protocols, achieving higher mAP scores for detectors like FCOS, RetinaNet, Faster R-CNN, and DETR, even with fewer epochs.

  • Characterising Decision Theories with Mechanised Causal Graphs: Characterising Decision Theories with Mechanised Causal GraphsThis paper addresses the question of how an individual's decisions should affect their beliefs about the expected outcomes they can achieve. It explores the influence of decision-making on perceptions of oneself, others, and expected utility calculations. The paper examines different decision theories, such as evidential decision theory, causal decision theory, and functional decision theory. The authors demonstrate that mechanised causal models can be employed to characterize and distinguish these decision theories, leading to the generation of a taxonomy of different approaches to decision-making.

  • LLM Censorship: A Machine Learning Challenge or a Computer Security Problem?: Large language models (LLMs) have shown impressive capabilities in understanding complex instructions, but concerns have arisen regarding their blind adherence to potentially malicious instructions. Existing defense mechanisms, like model fine-tuning or output censorship using LLMs, have limitations, as problematic responses can still be generated. This paper highlights the theoretical limitations of semantic censorship approaches, showing that it can be perceived as an undecidable problem due to LLMs' programmatic and instruction-following capabilities. The paper argues that censorship should be viewed as a security problem, and security-based approaches should be adapted to address potential risks from knowledgeable attackers reconstructing impermissible outputs from permissible ones. Reevaluating the problem of censorship with a security-focused perspective is necessary to address these challenges effectively.

Community Corner

Welcome to the Community Corner, our dedicated hub for interaction, innovation, and inspiration within the MetaLyst universe. Here's how you can get involved and make the most of your community:

  1. Your Voice, Our Evolution – We're committed to growing with you. Help us enhance your MetaLyst experience by providing your invaluable feedback. Desire a new section or more visuals? We're all ears. Let your thoughts shape our evolution. Give your feedback.

  2. Creators' Spotlight – The Metaverse is brimming with creators like you. From digital artwork to pioneering blockchain projects, your creations ignite our curiosity. Share your work with us and stand a chance to be featured in an upcoming edition of MetaLyst. Submit your piece.

  3. Partner with MetaLyst – Reach our dedicated and engaged readership by advertising with MetaLyst. Connect your brand with enthusiasts and professionals in the Metaverse industry. Fill out our quick partnership form to begin your journey.

Not a Subscriber? Sign up for the MetaLyst Newsletter to get the latest Metaverse news, views, and insights in your inbox every day.The Morning Brief

Join the conversation

or to participate.