LONDON - Jeremy Kahn is a senior writer at Fortune Magazine who is based in the United Kingdom. He writes about artificial intelligence (AI) and other disruptive technologies, from quantum computing to augmented reality. His weekly newsletter Eye on AI is one of the best non-technical sources of news and views about the technology. In this exclusive talk with The Yuan, he shares his thoughts on how transformational the latest AI systems are likely to be.
Is improvement actually happening exponentially?
There is no doubt that technology is not only improving, but doing so at an accelerating rate. As a journalist, this makes covering the sector both more interesting and more challenging, since it is hard to keep up with all the significant papers and product releases. In Kahn’s view, there also remains some controversy over whether the rate of technological improvement is in fact exponential, or actually something slower. Recent research from Stanford found that the token prediction capability of large language models (LLMs) like the generative pre-trained transformer (GPT) GPT-4 is improving at a linear, not an exponential rate.
Kahn also believes that observable advances in underlying technology sometimes have more to do with the interface between humans and machines than with the power of the tech itself. Even the AI model powering ChatGPT was not the most powerful one in existence at the time. Instead, the main reasons it proved to be such an eye-opener were that its interface was user-friendly and that the fine-tuning achieved with it by reinforcement learning from human feedback was highly effective.
Transformation by transformers
Now that the world has GPT models, with new ones being launching daily, how much impact will they have on the way thThe content herein is subject to copyright by The Yuan. All rights reserved. The content of the services is owned or licensed to The Yuan. Such content from The Yuan may be shared and reprinted but must clearly identify The Yuan as its original source. Content from a third-party copyright holder identified in the copyright notice contained in such third party’s content appearing in The Yuan must likewise be clearly labeled as such.