Keep pulling the thread on Benedict Evans.
OpenAI has publicly stated an expectation to develop AI models with human-level, PhD-level research capabilities by next year.
ChatGPT has between 800 and 900 million weekly active users.
Approximately 5% of ChatGPT's users are paying subscribers.
Despite having benchmark scores comparable to other leading models, Anthropic's Claude has very little consumer usage.
In terms of consumer usage for chatbots, ChatGPT is the leader, with Meta and Google's offerings having significantly less usage.
Benedict Evans believes OpenAI's market position is fragile because it lacks network effects, feature lock-in, a broad ecosystem, and control over its infrastructure costs.
Benedict Evans believes that neither Google nor OpenAI currently have the capability to reliably deliver the advanced, agentic product that Apple demonstrated for Siri two years ago.
Sam Altman has stated that current AI models have the capabilities of PhD-level researchers.
Demis Hassabis of Google DeepMind disagrees with Sam Altman's claim that current AI models have PhD-level capabilities.
Benedict Evans argues that the mobile platform shift was a key driver of Meta's growth, and the company would be much smaller without it.
Andrej Karpathy has stated on a podcast that he feels AGI is a decade away.
Mark Zuckerberg has stated that if Meta overinvests in AI compute capacity, the company can resell the excess capacity.