Microsoft is asking investors to temper their expectations for quick financial returns from Copilot, emphasizing the need to convince customers that the substantial monthly cost is worth it. While testers have mixed feelings about Copilot’s usefulness and price tag, Microsoft believes it justifies the price point in scenarios involving sophisticated information retrieval and task completion in Office 365, Teams, and Outlook. However, other portfolio parts, such as Excel and PowerPoint, are still in the learning phase. Microsoft is focused on helping customers realize the value of Copilot and building a business case based on time savings and productivity enhancements. They are optimistic that Copilot will be a profitable venture in the long term.
Elon Musk has open-sourced the code for his A.I. chatbot, Grok. This move is part of the ongoing battle between Musk and OpenAI over control of the future of AI. The controversy over open-sourcing generative AI has divided the tech world, with Musk advocating for transparency while others argue for guarding the technology. Musk aims to level the playing field by open-sourcing his code and compete with OpenAI’s popular chatbot. The model, weights, and architecture were released, but not the training data, under the Apache 2.0 license.
Per reports in the Information, OpenAI’s chatbot app store, launched by CEO Sam Altman, has had a slow start. Developers offering chatbots through the store have reported a lack of customers and limited support from OpenAI. Access to the store is currently limited to paying customers of ChatGPT, and developers are not yet able to charge for their apps.
Third-party plugins for OpenAI ChatGPT have been found to pose a security risk, allowing unauthorized access to sensitive data. Security flaws within ChatGPT and its ecosystem enable attackers to install malicious plugins without user consent, potentially hijacking accounts on third-party websites. OpenAI has taken steps to address these issues, including discontinuing the installation of new plugins. Additionally, researchers have discovered an LLM side-channel attack that exploits token length to extract encrypted responses from AI Assistants, highlighting the need for countermeasures to protect sensitive information shared in private AI assistant conversations.
The trite idea is that companies should focus on building and communicating clear business cases for their AI offerings. That isn’t wrong. Instead, I’d offer that being the guide for customers to examine, evaluate, and review AI business cases is the consulting opportunity to take advantage of.
Today’s stories cover a spectrum of business models. Will you make money on applications like CoPilot? That one looks broadly true. Or implementing models like Grok directly? Probably less so. Will there be money in app stores like OpenAI’s? I’ll offer this, which seems least likely. There isn’t one answer here, and that’s a feature, not a bug.