An Intelligent Global Brain

When I wrote The Global Brain, forty years ago, we were in the very early stages of computer networking. These early networks soon became the Internet, and later the Worldwide Web. From that came a whole plethora of innovations: streaming video, online shopping, news and banking, social media, the cloud, and more; not to mention the now ubiquitous website. However, it was generally just data, spread over a global network, one as complex as the human brain with exceedingly efficient search, but still just information of one kind or another. Now a whole new dimension is being added: AI (artificial intelligence). It has been there behind the scenes for several years, but is now becoming available to the everyday user.

A language processing AI called GPT-3, which can process this global data and summarize it in good English (and several other languages) is now catching the headlines in the form of ChatGPT—an app that anyone can access, ask questions, and get remarkably good answers. You can, for example, ask it write an essay about some specialist scientific subject, and be blown away by the results. Write a poem about some spiritual concept in the style of Shakespeare (say). Or find an error in your computer code. The possibilities seem almost endless. ChatGPT was only made public in November, but it has already taken the Internet by storm—within five days of its release it had a million users!

As well as a multitude of possibilities, it raises a number of issues: How will teachers be able to detect if a student’s essay is their own work or that of ChatGPT? Can we trust its answers (it sometimes makes mistakes)? GPT-3 was trained on the global data of 2021, so cannot give up-to-date responses. Currently, it is very expensive to operate; will that limit its development? Will it be integrated into search engines? Could it be hacked? Will it reinforce our socio-political biases?

Some of these issues may be resolved over time. Some may not. But here I want to explore some of the longer-term implications.

The next obvious step is adding audio so that we can speak to it and get spoken responses. The technologies of speech recognition and synthesis are well developed, and in common use with virtual assistants such Siri and Alexa. But whereas they currently just search the Internet for answers, now, with AI behind them, they’ll be able to process the data, and give more intelligent responses.

It will also bring an end to those, sometimes frustrating, human-operated call centers. No longer will we have to wait on hold, but will get immediate, intelligent answers—and in an accent we can understand.

ChatGPT wasn’t programmed with the answers it gives. It doesn’t actually “know” anything. GPT-3, on which it is based, is an AI that creates good English. But it does this statistically. It has been trained on a vast amount of English sentences and looks for the most likely word or phrase to follow a group of words. So if you ask “Who was the first person to walk on the moon?” it trawls its data, looking for the most likely continuation of the words “first person to walk on the moon…” and finds that “Neil Armstrong” is most likely. It is useful to know the way it works when you get weird answers of mistakes. But nevertheless quite impressive in its present scope. It will be much more so in the future. In addition, other apps are becoming available improving on ChatGPT..

As it becomes increasingly conversational, we may begin to feel there is a “person” there, not a human person of course, but our minds will project on to the AI the sense of there being someone there, and available to us at any time. Maybe we’ll feel we are each accessing the same virtual person. Where will that lead, I wonder?

Other forms of AI can also be accessed online. There are several images generators that will create excellent pictures based on simple descriptions. A number of video generators—although not yet high quality—and music generators. All of which will rapidly improve with time.

Beyond these foreseeable developments, online AI will bring many innovations that are probably as far beyond our imagination as the Internet of today was when the first personal computers appeared. And remember that the pace of development today is very much faster than it was forty years ago. In ten years’ time, AI will be much more powerful, perhaps surpassing the human brain in its capabilities. We will then be in a very different world indeed.

Welcome to the Intelligence Age!

Leave a Reply