I like grouping stories to see trends, and with Google doing their I/O event today and my expectation of a lot of AI stories coming from that, I wanted to clear the decks of my other AI threads.
First, let’s mention the Moat email. What’s that? It’s a leaked memo written by Google software engineer Luke Sernau that starts with “We Have No Moat. And neither does Open AI”. Referencing the idea of a defensive differentiation of products by having a “moat,” the memo highlights the quiet development of open-source algorithms, bringing innovations like language models on phones, multi-modality, and the potential for faster iteration on smaller models. A core concept here – owning the ecosystem of developers.
Also of note – news that OpenAI lost $540m last year and will need as much as $100b more funding, even after the $10 billion from Microsoft in January.
IBM announced Watsonx, an AI platform for business, which uses those open-source models and customizes them with specialties in domains and provides a toolkit for governance.
Microsoft 365 CoPilot is getting a further rollout. It’s now moving to an invite-only stage of 600 global customers, up from the initial 20. There are new features, too – bringing CoPilot to Whiteboard, adding DALL-E to Powerpoint, and adding CoPilot to Outlook.
Slack showed off an early version of its SlackGPT, which offers assistance to users on the tone or length of messages and can attend Huddles for a user and summarize them. No release date yet.
Think it’s not a big deal yet – the Pentagon does. Defense Department’s chief digital and AI officer said he worries about the profound havoc such tools could wreak across society.
“I’m scared to death,” about how people might use ChatGPT and other consumer-facing AI agents, Craig Martell said Wednesday.
There are two levels to my thinking on AI. First, there’s the tactical of knowing what’s happening to keep up with what the potential solutions to take to market are. Customers are looking to providers to answer the question of what’s possible. They’re also looking to answer why, and that’s the second level.
Is all this AI unleashed on the world a positive force? A quip in a podcast won’t answer that question, but the question itself clouds that it’s far more complicated than that. In some cases, yes; in others, no, and it will ultimately be complex and messy. Cumbersome and messy is also code for interesting and profitable with no single answer, which is prime space to build solutions beyond technical.