Is it already time to cite the Hype Cycle with respect to AI and my predictions for 2024?
First, Texas grades exams with AI, but refuses to call it “AI”. This seems to be the first step to using AI in administrative law. Whether or not, the grading itself is reliable or not, the whole process will for sure trigger some lawsuits.
Second, Devin, the AI software engineer was debunked. The admittedly impressive demo for the product launch was likely doctored. This acts as a reminder that large-language models (LLMs) are just a statistical way of recombining already-known knowledge on a very large scale. Indeed, the scale is so large that prices for storage hardware, even for good old HDD have risen by 20% since Q3 2023. At the same time, OpenAI continues to crawl the web and has reportedly hit a silly content farm. OpenAI is still trying to expand to new markets like Japan while the excitement about ChatGPT, Copilot and other tools seems to be decaying.
The other part of the AI boom has been computer chips. At least, innovations in this space are here to stay. The Bloomberg article about why making computer chips has become a new global arms race has some valid points.
Google continues working on its own “AI” chip. I found it very difficult to get my hands on Google Cloud Instances with NVIDIA chips. Maybe that’s why they intensify working on a competitive chip. If Google works on a chip, Meta has to follow, obviously. If they’re using the chip to detect and label AI-generated content on Facebook?
Top comments (0)