• Stability’s founder stated that models will run on mobile by the end of 2023, and their new Stable LM 3B can already do that – their strategy revolves around smaller, customizable models that open a range of new, practical, and cost-effective applications for genAI. - Stability
• “The next big breakthrough might not be the next version of GPT. It might be halving the cost,” says Microsoft expert, who believes that “Copilot on Windows is going to make everyone a power user,” and offers insights into Microsoft’s advantages in the AI race. - Semafor
• A few hundred $$ a month to run a GPT-3.5-powered feature – it doesn’t have to be extremely expensive to add AI features to your product, as long as you follow some principles (limit output of LLM, limit prompt size, stick to cost-effective models and not GPT-4). - Honeycomb
• Reducing inefficiencies in healthcare – an AI can predict missed care opportunities (when patients miss appointments, don’t schedule follow ups or are late) with up to 96% accuracy, it could improve patient outcomes and reduce physician burnout. - GE
• Visa will invest $100 million into commerce & payments genAI products – “Given the early stage of the industry we’d expect to make a range of smaller investments. That said, we have the ability to make larger investments where there is a strong rationale to do so.” - TechCrunch
• ToRA raises the bar for mathematical problem solving – combining LLM agents and external tools, it’s the first open-source model that achieves an accuracy exceeding 50% on the MATH benchmark. - GitHub
• Up to $90B valuation for OpenAI, $30B for Anthropic, $5B for Character.AI – some experts worry that we’re heading into overhyped bubble territory, these companies clearly have value but the valuations seem too optimistic at this time. - The Register, The Information, Bloomberg
