AI 服務暫時不可用，以下為來源正文，待恢復後補全翻譯。
It’s been interesting to see some people saying that the time for consumer AI is coming, meaning everyone outside the tech industry will start actually using AI agents the same way software engineers, product managers, designers, founders, and investors already do. It’s also around the same time that we’re getting the remarkable accomplishment of OpenAI publishing ten new results in mathematics and theoretical computer science, including several long-standing Erdős problems that expert mathematicians had been trying to crack for decades, produced by an internal version of their next model. When I think about those two things, the dissonance is quite incredible between how I, as a software engineer, and other people in the tech industry see AI agents, versus how little they’re actually used by other people and other professions. Part of the issue is that generative images, audio and video have kind of traumatised the general public. Many people’s first exposure to AI was negative, shaped by scary or depressing things like deepfake videos and cloned voices. From my conversations with people outside the tech industry, most people just saw that on social media, or really amateur early uses of AI in media, and thought, “This is AI.” And I say this as someone who contributed to the problem by working for an early-stage AI image startup in 2024. Meanwhile, all these incredible advancements in science and research are being powered by LLMs, but much of the general population still sees AI as undesirable. I think that’s why adoption has been slow. Most people are still stuck on that first impression, and in hindsight, it was probably a mistake that we made generative media so scalable before we were actually ready for it. And I mean just generative media. The invention of ChatGPT, and all the power AI agents are creating in so many areas, has been incredible to experience firsthand. But yeah, there’s a rebranding effort needed now to get people genuinely excited about AI actually helping them.