Members

Open banking has not lived up to its potential in Great Britain, according to a report by the Financial Times. While it has seen success in some areas, such as faster lending decisions, it has not reached profitability, with higher interest rates cooling investor enthusiasm. Open banking technology allows customers to share their financial information with other banks, apps, and online retailers, and allows lenders to permit pay-by-bank remittances without card intermediaries. The excitement around open banking accompanied the past decade’s U.K. FinTech boom, making London a leader in the sector. However, many users are unaware of the technology’s availability and do not see the benefits of digital wallets like Apple Pay. The awareness gap for open banking is not just confined to the U.K., with 56% of American consumers not familiar with pay by bank. However, incentives can help bridge this gap, with Generation Z and high-income individuals being particularly receptive to pay by bank, especially when coupled with incentives.

Read Article

Researchers from Meta’s FAIR team and The Hebrew University of Jerusalem have discovered that forcing large language models to “think” less actually improves their performance on complex reasoning tasks. The study found that shorter reasoning processes in AI systems lead to more accurate results while significantly reducing computational costs. The researchers discovered that within the same reasoning task, “shorter reasoning chains are significantly more likely to yield correct answers — up to 34.5% more accurate than the longest chain sampled for the same question.” This finding held true across multiple leading AI models and benchmarks. Based on these findings, the team developed a novel approach called “short-m@k,” which executes multiple reasoning attempts in parallel but halts computation once the first few processes complete. The final answer is then selected through majority voting among these shorter chains. The researchers found their method could reduce computational resources by up to 40% while maintaining the same level of performance as standard approaches. “Our findings suggest rethinking current methods of test-time compute in reasoning LLMs, emphasizing that longer ‘thinking’ does not necessarily translate to improved performance and can, counter-intuitively, lead to degraded results,” the researchers conclude.  The study points toward potential cost savings and performance improvements by optimizing for efficiency rather than raw computing power.

Read Article

While EMV 3DS has many benefits, adoption may be slow in the regions where EMV 3DS is not mandatory. Reasons may include the following: Data inconsistency. The quality of merchant data provided in EMV 3DS plays a critical role in issuer fraud detection. Merchants may be reluctant to share more data and may decide to provide a minimum set of data elements excluding optional data elements. There are cases when the data provided is not accurate, causing issues in fraud engines. Approval rates and cardholder friction. Shopping cart abandonment has been one of the major reasons that EMV 3DS adoption is low. Many enhancements have been added to the protocol from EMV 3DS 1.0 to EMV 3DS 2.x to challenge the cardholder only when needed. Complexity of integration. EMV 3DS integration is complex and adds an additional layer of authentication flow before authorization, resulting in higher implementation costs. Most systems are built with synchronous authorization request and response; EMV 3DS is a major change since it requires an asynchronous transaction processing flow with multiple steps. Liability shift. EMV 3DS is designed to help with fraud. However, determining if, how and when a liability shift occurs for merchants is not a simple answer and depends on several factors. Payment network and local regulatory requirements should be checked for specific use cases to assess any applicable liability shifts. Some factors are: Region and payment network. It is important to be familiar with payment network rules for EMV 3DS usage. Merchant category code (MCC). Not all MCCs are allowed for a liability shift.

Read Article

Local Falcon’s research showed that AI chatbots often bypass traditional SEO metrics such as link volume or page ranking, relying instead on relevance, prominence and authority. That means brands that surface in AI responses are those that offer “citable evidence” of expertise, such as real data, FAQs, testimonials and contextual content that AI can easily access, the study said. AI chatbots don’t prioritize traffic or backlinks when they choose which brands to list in their responses to consumers. Instead, they look for provable expertise. The good news is that smaller brands that are expert in what they do can leapfrog over well-trafficked retail sites in search listings. For example, stores shouldn’t just say they are the leading retailer in a niche product but actually prove it by explaining why they’re the best. It also helps for the brand to be mentioned in places like social media. For smaller businesses, it’s an opportunity to compete with larger players — if they provide genuine value and clear proof of expertise. “It gives smaller stores a shot at being found if they play it right and they’re really good at what they do,” David Hunter, CEO of Local Falcon said. Hunter thinks that Google transitioning from traditional search to AI-powered search is the cause of this surprise finding. Ultimately, location will continue to play an important part in search even for an AI chatbot. Location-based marketing and messaging is a powerful tool that drives higher engagement, foot traffic and revenue, according to Radar CEO Nick Patrick. Location is important not just for retailers but also for financial services — to do things such as find bank branches, real-time fraud detection and geo-triggered cash back promotions through retail partners. But instead of location, the AI appears to favor authoritative results from sources including social media, Reddit, and other community forums, which were previously not prioritized by traditional algorithms.

Read Article

Odyssey is taking a different approach than many AI labs in the world modeling space. Odyssey has developed an AI model that lets users “interact” with streaming video. It designed a 360-degree, backpack-mounted camera system to capture real-world landscapes, which Odyssey thinks can serve as a basis for higher-quality models than models trained solely on publicly available data.  The model generates and streams video frames every 40 milliseconds. Via basic controls, viewers can explore areas within a video, similar to a 3D-rendered video game. Powering this is a new world model, demonstrating capabilities like generating pixels that feel realistic, maintaining spatial consistency, learning actions from video, and outputting coherent video streams for 5 minutes or more.” World models could one day be used to create interactive media, such as games and movies, and run realistic simulations like training environments for robots. “Interactive video … opens the door to entirely new forms of entertainment, where stories can be generated and explored on demand, free from the constraints and costs of traditional production,” the company says. “Over time, we believe everything that is video today — entertainment, ads, education, training, travel, and more — will evolve into interactive video, all powered by Odyssey.” The model can currently stream video at up to 30 frames per second from clusters of Nvidia H100 GPUs at the cost of $1 to $2 per “user-hour.”

Read Article