STOP ACTING LIKE YOU'RE FAMOUS
Focus on what makes you enjoy your leisure activities and stop optimizing for a non-existent audience.
Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Focus on what makes you enjoy your leisure activities and stop optimizing for a non-existent audience.
Sponsored
The #1 concern around GenAI today is accuracy - how can enterprises scale a technology they can't trust? Narratize is cracking the code of GenAI accuracy for every innovative brand. Narratize is built to deliver accuracy, always:
There have been reports of the open source pioneer Mistral seeking several hundred million dollars of funding to train more models.
Stability AI has made its latest text-to-image AI model, Stable Diffusion 3, available to some developers via API and its new content creation platform called Stable Assistant Beta. The model is still in preview and not yet available to the general public.
Sanctuary AI will deliver its humanoid robot, Phoenix, to a Magna auto manufacturing facility, following Magna's 2021 investment in the company. Phoenix is notable for its walking capability and dexterous hands aimed at enhancing manufacturing efficiency. The specifics of the pilot, including the number of robots and deployment details, remain undisclosed.
The new base Mistral report was released today. Additionally, their own instruction tuned model was released under a similar permissive license. They report a strong MMLU and HumanEval performance along with incredible multilingual English, French, Italian, German, and Spanish performance, function calling, and 64k native context length.
The latest completely open 7B language model had an upgrade with a much-improved reasoning score on MMLU. Its development team found that changing the data mixture had a substantial effect on performance. They have provided the exact statistics of new data sources and percentages that led to this improvement.
This paper describes Stability AI's diffusion transformer model for audio synthesis.
Sponsored
The third annual Semantic Layer Summit [14] is less than a week away! This free, one-day virtual event features some of the top voices in enterprise analytics, BI, and genAI — including leading technical leaders at Mode, Netflix, T-Mobile, and AtScale.
A deep dive interview from industry experts on the challenges and solutions of getting AI models into production and where MLOps differs from traditional engineering. They discuss why so few ML projects make it to production and how to focus as an organization to actually launch.
The Effort library allows for real-time adjustment of the number of calculations performed during inference of an LLM model, resulting in significant speed improvements while retaining most of the quality. Despite some implementation overhead, initial results suggest that the Effort library has the potential to significantly improve LLM inference speed while maintaining quality. The author encourages others to test the 0.0.1B version and provide feedback to further improve the library.
This study revisits using diffusion models for image-to-text generation. It introduces the LaDiC architecture, which enhances diffusion models' performance in image captioning tasks.
Wayve's Lingo-2 is a VLM train in simulation for autonomous driving tasks. It performs driving actions based on video input. Lingo-2 includes descriptions of reasoning for the driving actions it takes based on the scene.
Researchers are developing alternative architectures to address the limitations of Transformers in long-context learning, generation, and inference speed, showing competitive quality at smaller scales but uncertain scalability. The rapid progress in this subfield suggests that the Pareto frontier will continue to expand, enabling longer context modeling and greater inference throughput, ultimately increasing the number of use cases for AI.
Building a successful AI company requires navigating complex short, mid, and long-term dynamics and maintaining elite speed and execution, owning more of the stack, gathering unique data, and leveraging synthetic data generation. As the AI industry matures, companies must adapt to shifting talent dynamics and understand the machine they are building and the axes of competition they are predicated on to create lasting moats and stand out amidst the noise.
Verifying training when sharding across devices can be challenging - this handy snippet verifies that things have been done and updated properly.
Luminal is a deep learning library that uses composable compilers to achieve high performance.
Google Maps will use AI to summarize directions to EV chargers as well as provide reliability and wait times.
Senators Kirsten Gillibrand (D-NY) and pro-crypto Cynthia Lummis (R-WY) released the Lummis-Gillibrand Payment Stablecoin Act to regulate the stablecoin industry. The act would require stablecoin issuers to hold one-to-one cash or cash-equivalent assets and prohibit unbacked algorithmic stablecoins. If an issuer experiences insolvency, there would be an FDIC conservatorship and resolution. There are some concerns that the bill is too strict with its definition of cash equivalents and that it could prohibit innovation with new stablecoin designs.
BlackRock recently released a video about its Bitcoin ETFs, calling Bitcoin “the most widely adopted digital asset in the world.” The asset management firm's US Head of Thematics and Alternative ETFs, Jay Jacobs, noted the cryptocurrency's immense status and promise in a video published this week.
Kraken has launched a self-custodial wallet, allowing users to manage NFTs and tokens and conduct DeFi transactions across 8 blockchain networks. The Kraken team published the app's open-source code on GitHub for enhanced security through third-party review. The wallet is considered a hot wallet due to current iOS limitations. Kraken is committed to security and transparency.
Worldcoin is set to launch its own Ethereum Layer 2 blockchain, "World Chain", to improve its ability to onboard new users. The Layer 2 is designed to prioritize human-verified users, with benefits like priority block space and free gas. World Chain is expected to be released this Summer.
The Runes protocol, set to launch in April 2024, is a new fungible token standard on Bitcoin that aims to improve upon the shortcomings of its predecessor, BRC-20. By integrating with Bitcoin's existing UTXO model and utilizing OP_RETURN for data storage, Runes minimizes on-chain footprint and network congestion while offering a more user-friendly and accessible token creation process.
Modularity breaks up blockchain functionality into three parts — data availability and consensus, settlement, and execution — each of which gets its own chain or environment. Prominent projects exist in each sector, along with interoperability to unite liquidity, with the goal of integrating together to make something greater than the sum of their parts. The commoditization of block space and execution will enable applications to operate at much larger scales with grander visions.
Faraway, a gaming studio, has acquired the Legends Of The Mara and HV-MTL brands from Yuga Labs. It plans to integrate them into Faraway gaming products while maintaining close ties to Yuga and Otherside. Spencer Tucker, Yuga's Chief Gaming Officer, will move to Faraway as Chief Product Officer to ensure a seamless transition.
Legendary cryptographer Wei Dai weighs in on the pros and cons of whether a base layer like Ethereum should offer privacy by default. On one hand, it is technically feasible for privacy to be implemented, and having it as the default option makes it much more normalized, especially for users from traditional finance. On the other hand, this would be incredibly complex to implement, remove some level of neutrality, and not align with the rollup-centric roadmap's idea of end users never interacting with the base chain. Dai leans towards a transparent base layer with L2s and apps leading the experimentation with privacy based on their goals.
Venom is a Threaded Virtual Machine (TVM) with features like account abstraction and dynamic sharding to improve transaction speed and reliability.
The game integrates tower defense and monster collection gameplay and is a collaboration between Gravity and Ronin's developer Sky Mavis.
Crypto adoption will shift from buying tokens to stake and trade to earning them through new user experiences like gaming.
Nansen explains how to use its onchain data to analyze tokens, using PENDLE as an example.
Virgin Wines partnered with Norwich agency Borne for its first major rebrand in 20 years, unveiling a new logo and color palette aimed at highlighting the joy of wine and enhancing perceptions of expertise and quality. The swirling logo, inspired by wine in a glass, features gradient colors representing various wine types and reflects Virgin Wines' commitment to authenticity and premium experiences.
Figma announced Code Connect for developers, typography and gradient variables, and the Library Analytics API during its Framework event, which focused on design systems. Code Connect, a Dev Mode tool, imports production-ready code snippets directly into the platform. Typography and gradient variables give designers new levels of control, scalability, and accessibility. The Library Analytics API enables enterprise customers to store and analyze library data outside of Figma.
Last year, Google launched Demand Gen campaigns to help advertisers increase demand using visual storytelling. New generative image tools make it easy for creatives and advertisers to create stunning, high-quality images in a few steps using prompts. The new capabilities will optimize the testing of creative concepts, whether you're experimenting with different types of images or building your own from scratch.
Lack of communication, inadequate documentation, and an outdated mindset are common design handoff problems. To avoid developer frustration, work with them on the best documentation format, align to the development framework, give them more context, and provide onboarding for navigating your files. Maintain open communication channels, don't over-rely on tools, and structure your handover process in line with your reality.
Quality decision-making requires building a shared mental model among those who make those decisions. An organization's semantic environment is determined by its beliefs, experiences, and preferences. However, these should be checked against users' reality, or your company will decide based on partially selected data to support a particular viewpoint.
A collection and documentation of what AI looks like in UI, its core interactions, and patterns.
Onlook empowers designers to streamline the product development process while saving countless hours for their team.
Henosia connects directly to your team's code repository to visually edit and preview code for designers.
Kobo's new color e-readers, the Libra Colour and Clara Colour, feature E INK Kaleido 3 screens, which offer 16 grayscale levels and 4,096 colors, enhancing reading experiences. Retaining traditional e-reader benefits like long battery life and lightweight design, color e-readers are ideal for reading textbooks, highlighting text, making notes, and creating illustrations, making them more versatile than traditional e-readers.
Up is an Australian mobile-only digital bank with a mission to reinvent banking while making it more fun. Up revolutionizes banking by challenging traditional design assumptions and adopting a user-centric approach to design inspired by popular apps. It promotes better financial literacy through streamlined onboarding and accessible, modern security measures.
Designing strategically is usually about considering the big picture, understanding overarching goals, conducting tests, and learning as you go. The broader scenario requires empathy, which can be developed by talking to stakeholders and department leads about the project objectives and context. If you are hired for a high-level strategic project, you must make trade-off decisions related to business strategy.
You can improve product value without adding unnecessary complexity.
Are we answering the right question? What's the simplest version of this? What's the most magical version of this? What would get somebody to talk about this?
Even in tough times like these, growth design projects can and should still be ethical and provide users with real value.
Sponsored
Passwords cause friction, and friction kills conversions. Passkeys let users log in with the method they already use [4] to unlock their device.
The “Dead Internet” theory not only suggests that AI-generated content is replacing human-created content online, but that this was purposely orchestrated by governments and corporations to control public perception. It was once dismissed as a conspiracy, but the theory has been gaining credibility again as AI advances. AI has been flooding social media platforms with what's referred to as “AI slime”. This influx of AI-generated content poses challenges for users and advertisers, who may struggle to discern authenticity and engagement.
YouTube has introduced targeted ad options for Shorts, empowering advertisers to reach audiences interested in specific topics like sports, gaming, food, and fashion. This move enhances ad customization and provides advertisers with greater control over their campaigns.