A new internet phenomenon has taken the world by storm. ChatGPT has managed to get 100 million subscribers in a far shorter period than any other App ever launched on the internet. The reason is clear. It has revolutionized the search function on the web. ChatGPT uses clever Artificial Intelligence (AI) algorithms to come up with human like answers to any question you might have. You can ask ChatGPT which small car would be the most suitable for yourself. You could also ask for a suggestion on where to spend the next family beach holiday. You can even request something much more complex and comprehensive, such as writing a short story about a topic in the style of Hemingway.
Instead of giving a list of websites which might contain the answer you are looking for, it converses and gives a very authoritative answer.
The applications are plenty, the benefits seem endless (hence the many copycats). Most students probably love it because they spend minutes rather than hours doing their assignments. It will be used to replace call centers (however if you have that unique query you will still be stuck with no answer and no one to ask). I can also imagine that professional fields like law, consulting and medicine can be streamlined using chat bots resembling ChatGPT. Because it uses mathematical formulas to take a set of information to predict the next set of information the technology could be used in the Pharma industry to try and predict new viral outbreaks or even formation of new viruses.
In this excitement of a reimagined search functionality, it is easy to forget about the darker side. Three flaws stand out.
Firstly, a system that can write code could surely be taught how to write its own code for itself to change itself on a constant basis. Imagine regimes like North Korea with their enormous army of programmers. They might use the technology to write destructive computer viruses that constantly change and learn form the way it is being countered. This is an immense destructive force that will cause chaos from anything like airports to power generators.
Secondly, these AI systems rely on the information available. In the case of most search chatbots, it relies on the information available on the internet. If a conversation with your computer chatbot becomes the dominant form of getting information, who will bother to publish new information if that would only be “stolen” by the Chatbots. Regulators should force the internet search companies to cite the pages where the Chatbot got the information used from.
Thirdly and most importantly, the AI revolution has been hailed as a boon for productivity. That might be so. Countries like the UK are desperately looking for an advantage to make their economy more productive. Integrating these AI functionalities are going to have a similar affect to the introduction of personal computers. The trouble is that even though we have seen some real productivity gains over the past 30 years, it coincided with an increase in the Gini Coefficient worldwide. The real beneficiaries of that gain in efficiency are those with capital. They will get a higher productivity per unit capital deployed. Yet a stable socio-economic and political environment can only exist with a broad middle class.
The development AI chatbots reminds me of the early days of social media, where the quest to find long lost friends was so strong that any possible shadier sides would be ignored. In Hindsight we were gullible. Very few other inventions have caused more harm to the human race in so many forms as has social media. From a rise in depression among the youth, to online bullying and attacks on the personal integrity – all because of social media. It didn’t stop there though. Fake news caused our societies to be more divided than ever. Governments, like that of Myanmar used social media to co-ordinate attacks of the Rohingya. Minority views were made to seem like the majority view, and majorities thought they were in the minority. Most were afraid to speak out. When Facebook and the likes were created, no one thought that they could cause to harm anyone. Yet they have. So will AI chatbots.