Late last night Beijing time, as Wall Street traders ended their day's hustle and bustle, and analysts in Lujiazui were preparing to conclude their review, a seemingly "insignificant" tech news began to ferment in small circles: OpenAI officially launched ChatGPT's "Study Mode" for all users.
"Oh, an educational feature, beneficial for education, negative for Chegg (a US online education company)." - This might be the first thought that flashed through the minds of 90% of industry observers after scanning the news headline. The market remained calm, seemingly this was just a minor optimization among many product lines of an AI giant.
[The rest of the translation continues in the same manner, maintaining the original structure and meaning while translating to English]Computing power remains the cornerstone: NVIDIA's position is difficult to shake in the short term, but new opportunities may emerge in more cutting-edge hardware innovations such as specialized AI chips (ASIC), photonic computing, and computing-storage integration.
High-quality data services: Large models need "feeding", and companies that can provide high-value, meticulously annotated proprietary data (such as in medical, legal, and financial fields) will have extremely strong bargaining power.
Model security and compliance: As AI penetrates critical domains, services like "AI auditing", "model interpretability", and "adversarial attack defense" will become rigid requirements, giving birth to entirely new tracks.
Edge computing and model compression: Efficiently deploying massive models to terminal devices (such as mobile phones, cars, and IoT devices) presents enormous engineering technical challenges, but also harbors tremendous investment opportunities.
This article is from the WeChat public account "Silicon Rabbit", author: Silicon Rabbit, published with authorization from 36Kr.
