Google has apologized for the rollout of AI Overviews in a blog post. The company attributes the inaccuracies to “data voids” and people asking unusual questions. Google is working to fix the issue by limiting AI Overviews for nonsensical queries and satire. The introduction of AI Overviews has been a PR blunder for Google, as it competes against OpenAI and other AI search startups.
OpenAI is launching programs to make ChatGPT more accessible to schools and nonprofit organizations. ChatGPT Edu is designed for universities, offering improved multimodal capabilities and enterprise-level security. Nonprofits can access ChatGPT Team at a discounted rate, and larger organizations can get ChatGPT Enterprise at a 50% discount. These initiatives come amidst recent resignations and controversies at OpenAI.
Free ChatGPT users now have access to custom GPTs, data analytics, chart creation, and other features previously available only to paid subscribers. While they can use existing custom GPTs, they cannot create their own. Paying subscribers still have the advantage of fewer message limits. The controversial Sky voice has been removed, but other voices are still available on ChatGPT.
All this investment in AI may pay off just for performance– Microsoft conducted benchmarks comparing its new Surface Laptop to the MacBook Air M3, showcasing better performance in Cinebench and Geekbench tests. The Surface Laptop outperformed in a HandBrake ToS benchmark and demonstrated longer battery life in web browsing and video playback simulations. Additionally, Microsoft highlighted the superior NPU performance and efficiency of its Snapdragon X Elite processor. These benchmarks provide insight into the upcoming launch of Microsoft’s Copilot Plus PCs.
According to Gartner, revenue from AI semiconductors is projected to reach $71 billion in 2024, with a 33% increase from 2023. AI server accelerators are expected to reach $21 billion in value by 2024 and $33 billion by 2028. AI PC shipments are forecasted to reach 22% of total PC shipments in 2024, and by the end of 2026, all enterprise PC purchases will be AI PCs. AI chips revenue from computer electronics is projected to account for 47% of total AI semiconductors revenue in 2024.
Why do we care?
Even if the AI boom doesn’t result in software advancements, the advancement of processors will still be significant. I wanted to note Google and OpenAI’s missteps here – Google for obvious reasons and OpenAI for not finding the balance of paid versus free in their freemium model. That said, note they are targeting sectors too, and this is not the first time we’ve covered the nonprofit space recently.