13 Sep, 2024

OpenAI Hoping For A Big Leap in AI Reasoning

Imagine an AI that doesn’t just answer your questions—it thinks deeply before responding, revolutionizing how we approach complex challenges. OpenAI’s new o1 model family represents a bold leap forward in AI reasoning, and it's now available to ChatGPT Plus subscribers.What Is Chain-of-Thought Reasoning?Unlike previous models that respond instantly, the o1 models take their time, working through problems methodically. These models use what's called 'chain-of-thought' reasoning. Think of it as an AI working through a problem step by step, like a detective piecing together clues to ensure the most accurate answer.Why This Matters for STEM, Coding, and ResearchFor professionals in STEM fields, coding, and academic research, these reasoning capabilities can be a game-changer. Whether you're analyzing complex data sets, solving intricate equations, or coding applications, the o1 model promises better accuracy and smarter solutions. Even the smallest version of this model surpasses OpenAI’s previous top-tier models like GPT-4oExciting Features of the OpenAI o1 ModelOne of the most exciting features of the o1 model is its ability to think ahead and plan. OpenAI shared videos showcasing its abilities, from coding video games based on a single prompt to solving intricate logic puzzles that rival PhD-level challenges in biology, chemistry, and physics.Limitations on Non-STEM Knowledge and Future UpdatesWhile OpenAI’s o1 excels in reasoning tasks, it’s not without limitations. For now, the model’s factual knowledge on non-STEM subjects, like trivia and biographies, remains on par with smaller models like GPT-4o-mini. OpenAI has promised to improve these aspects in future releases.What’s Next for AI Reasoning?Training AI to think before it acts has the potential to unlock vast benefits—but does this heightened intelligence introduce new risks? How will this model reshape industries reliant on fast, accurate reasoning, such as healthcare, finance, and education?Deliberate Reasoning vs. SpeedSo, what do you think—should AI be more deliberate in its reasoning, or is speed more important? Share your thoughts on how the o1 model could revolutionize STEM, coding, and more!

09 Sep, 2024

Japan’s Big 3 Banks to Use Stablecoins, Swift for Cross Border Payments

Japan’s big 3 banks, MUFG, SMBC and Mizuho are involved with with a cross border payment system, Project Pax, that aims to use stablecoins instead of correspondent banks. However, in order to ensure that corporate customers can trigger trade payments in the conventional manner via their banks, Swift payment messages will be integrated with the stablecoin system, which is based on Progmat. Hence, the clients don’t touch the stablecoins.Given the efficiencies of stablecoin payments, the aim is to support 24/7 payments to address the G20’s goals for faster, cheaper, more transparent cross border payments. The total market capitalization of stablecoins has reached $170 billion with cryptocurrency as the primary use case, although they are also used for P2P payments.Meanwhile, Progmat is a DLT-based tokenization network founded by MUFG, in which Mizuho and SMBC are joint venture partners alongside other institutions. Additionally, there’s a Progmat Coin solution which is primarily used for trust-based stablecoins, which have ring-fenced assets.Progmat has been working with Datachain for some time on interoperability solutions to connect the permissioned Progmat network that uses the Corda enterprise blockchain with various permissionless networks. Those include Ethereum, Avalanche, Polygon and Cosmos. Additionally, Datachain founded TOKI a cross chain bridge that will run DeFi style liquidity pools for exchanging different currency and blockchain stablecoins. These solutions will be used as part of Project Pax, which is a joint venture between Progmat and Datachain.Given the work that’s already been done, they aim to start testing a prototype ‘shortly’. The Pax members also plan to collaborate with other financial institutions in Japan and elsewhere, looking to go live by 2025.Regarding the Swift integration, the messages will instruct payments to settle via blockchain networks. This is seen as addressing bank operational issues and removes the need for corporates to deal with the unfamiliar concepts of wallets and keys. Plus, it addresses AML and other compliance concerns.Other stablecoin and DLT initiativesProgmat already has several collaborations for stablecoin payments. Last year MUFG and DRW Cumberland started exploring JPY and USD stablecoins for institutional crypto settlement. This will involve the Japanese wallet provider Ginco issuing XJPY and XUSD stablecoins.It also has a partnership with existing Japanese stablecoin JPYC and with Standage, a Japanese specialist in export settlement.Stepping back, other DLT initiatives are targeting cross border payments. Of course, there is Ripple On Demand Liquidity solution that uses XRP as an intermediate currency and is used by remittance companies.On the institutional side, JP Morgan, DBS Bank and Standard Chartered are involved in Partior that uses DLT to smooth the correspondent banking payment process. This enables banks that connect to the network to make cross border payments 24/7. While the client banks can make instant payments, the settlement banks still have to settle up between themselves, which they do in the conventional manner. Both SMBC and Mizuho have done some work with Partior.More recently, the BIS launch Project Agorá which has similar goals to Partior. Its major advantage is the involvement of seven central banks that can provide wholesale CBDCs. This would sidestep Partior’s issue of the settlement banks needing to settle between themselves conventionally.Project Pax is one of an ever expanding range of projects in the realm of stablecoins, tokenized deposits and DLT payments.