My personal learning curve has been steep over the past few weeks since ChatGPT became available to the public. And there is so much more to try out. But as things are moving at an almost unbelievable pace, here is an initial list of thoughts and brief analyses.
You are very welcome to join in and sing along:
It’s the End of the World as We Know It (and I Feel Fine).
Famous song by R.E.M.
With the release of ChatGPT and the image generator DALL-E, both from Open AI, artificial intelligence has been put on the map. It will turn an incredible amount of things upside down. And it will have effects comparable to the Industrial Revolution or the creation of the Internet.
AI is not going away. It will be with us for the rest of our lives. Whether we like it or not. So let’s figure out how to deal with the new situation. Both in terms of the overwhelming possibilities and the fault lines in terms of ethics, democracy or security.
Everything happens super-fast right now. It is almost impossible to keep track. I had to realize that within a few days ChatGPT had gone from “Wow, it can write a sweet birthday poem to my niece” to “Oh my God, it has passed the bar exam, the exam for American doctors and got a Master of Business Administration!”.
What will ChatGPT do next? Build a nuclear power plant without nuclear waste? Run for president?
AI saves time and increases labour productivity per person many times over. So AI has the potential to transform the entire global economy. How much faster will everyone be? Imagine integrating an AI like ChatGPT into the legal system, into medicine, into science, into finance. Into media, into art (already happened), into gaming. You name it. And this is probably happening all at the same time.
Sounds crazy? It is not. I think Yuval Harari mentioned in one of his fantastic books that all the world’s trade in the 15th century could be done today with four or five large container ships. In other words, a person today is perhaps a thousand times more productive on average than a hundred years ago. What will happen if, in the next few years, AI increases that productivity by another hundred times? To be honest, the idea of such acceleration makes my head spin.
If you don’t try it, you don’t know what it’s like. One is to philosophize about ChatGPT at the coffee table, the other is to dive in and observe what AI does to you. You must feel it. The whole experience right now is so amazing, so astonishing – and terrifying at the same time.
But why ChatGPT is such a powerful tool, it takes a while to learn how to use it. At first glance it looks very simple, but if you want to get a result of substance, you have to try different things to find relevant prompts.
I am superman now! And you too! Think of Google Maps. It gives people like me, who occasionally get lost in the Old Town’s maze of alleyways, the superpower of being able to navigate home safely with my smartphone. In most places in the world.
AI now gives the average person many additional superpowers. For example, ChatGPT now allows me to write essays on topics I haven’t the faintest idea about. “Hey Chat, could you please write me a short essay on architectural styles in Sweden in the 19th century?” Or ChatGPT will quickly write a message for the bulletin board that would otherwise have taken me 15 minutes. And suddenly I can have ChatGPT write code using natural language commands. A very special superpower. Not to forget the image generators like DALL-E, which give me the superpower of painting and illustrating, turning my dreams or even nightmares into a finished image in a few seconds.
ChatGPT’s answers always sound very plausible. Even if the answer sometimes is nonsense at second glance. The AI writes grammatically perfect paragraphs, but it’s the depth and the reflection of the content that still isn’t there. And, as far as I know, AI does not understand irony. So, it is never funny on purpose, only by accident.
As a result, one could argue: ChatGPT is not ready for commercial scale. But this is just the first (public) version. The next version of ChatGPT is supposed to come out this year. People say it will be trained on some hundred times more data. And it will probably be much more sophisticated in terms of how many kinds of different things it can do. If that is the case, discussions about commercial use would immediately intensify.
Good questions lead to good answers. So, anyone who has learnt to write precise briefings has an advantage now. Suddenly, working in marketing was good for something, ha ha. The ability to write scripts for movies or plays with appropriate instructions certainly helps too. Systemic thinking anyway. And my initial experience is that people who are familiar with the structure and composition of journalistic texts get there quicker.
I have found some prompts that work for my business. I am beginning to understand, at least a little, how to communicate with ChatGPT, what to type into the text field. But I have no idea about the power behind it. ChatGPT is a black box. It is not an explanatory AI. You put something in, you get something out. But the “why are you giving me this answer and not another” remains a mystery.
To be continued. There is so much more to discover.