My thoughts on ChatGPT

I have been using ChatGPT quite often over the past couple of months, especially for work-related questions, and I must admit that I have been impressed with the answers and insight the tool has provided.

Yesterday March 30th 2023, the Italian Privacy Authority (Garante della privacy) took the decision to temporarily suspend the ChatGPT service due to several reasons (https://www.garanteprivacy.it/home/docweb/-/docweb-display/docweb/9870847)

My big boss Gennaro Scalo asked me for an opinion on the subject (https://www.dhirubhai.net/feed/update/urn:li:activity:7047615549874335744?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7047615549874335744%2C7047643283220525056%29 ), which gave rise to this article.

I have read the Italian Privacy Authority's statement, and to be honest, I agree with the reasons behind this decision.

The main aspect is that there is no clarity about the data sources and how information is stored and there are no filters for young people of any ind. Above all, this decision was triggered by a data breach of past March 20th that disclosed a lot of information about the uses that had a paid subscription. So, from a legal point of view, the decision was probably inevitable.

Over the past few weeks I thought a lot about this generative AI technology, so I take this opportunity to share with you some thoughts about ChatGPT.

As a technologist, as I said, I am really impressed!

I used ChatGPT to generate source code snippets, and in most cases the provided answers where basically perfect and aligned with the directives I provided. I also used ChatGPT to know more about very specific risk quantification topics and, again, the answers were very good. The process is iterative, meaning that the first answer is typically high level, and then we can as more and more details. I also asked questions about specific strategic aspects and again I got very interesting answers. Overall, I realised that ChatGPT is an incredible time-saving opportunity.

From the other side, on a personal basis, I must admit that over time, I have developed a growing concern and for several reasons. I report below the main 3.

1)????ChatGPT IS ADDICTIVE!

Recently a started studying a bit of neurosciences and now I understand why. When you use ChatGPT you feel you have the power to quick get the information you have been looking for. So, after you make the question, you immediately get the answer, that islike an award. In neurosciences, “quick time”, “ease of access” and “award” are the basic ingredients to trigger addiction, since when they are put together, the brain will generate dopamine which known as the “hormone of pleasure” and this in turn will require to ask more. ChatGPT is super easy to access and quickly provide very good answers, so… it’s addictive.

2)????TRUST CAN BECOME AN ISSUE

As I use ChatGPT and I consistently get good answers, my trust on the tool increases. After all this is just what happens with people, right? Initially I can be skeptical, and I will use other sources to make sure the information I get is indeed correct and trustful, but over time I will stop doing these double checks and I will simply trust ChatGPT. This is another of the concerns raised by the Italian Privacy Authority, as there is a huge potential for "algorithms" to steer people's opinions. This is nothing new, as social media has been doing this for years (watch the docufilm "The social dilemma").

3)????THINKING PROCESS

This is definitely the worst and most dangerous, although the timeframe is probably decades-long. So, we have a tool that is easy to access, responds quickly and that we trust. In both professional and personal life, we constantly need to know more about specific topics since we need to take decisions about what to do. What do we do? Easy, we ask ChatGPT, because we know we can quickly and easily get reasonable answers we can trust. By repeating this process hundreds or thousands of times, we will eventually end up delegating our thought process to ChatGPT! And this is, of course, bad, crazy and very dangerous. If you think that social media already started this process… well, I do agree. Thanks to my manager Nikki Mills who pointed me to this book " https://www.amazon.com/Arguments-Deleting-Social-Media-Accounts/dp/125019668X" and yes, I admit I asked ChatGPT to provide me a 1000 words summary of the book :-( ...

I think that the real reason the Italian Privacy Authority made the decision to suspend ChatGPT is that they understood how powerful and potentially dangerous this tool can be and the risks it can pose.

Progress cannot (and must not) be stopped, but it must be properly regulated to avoid potentially disastrous effects. And it seems that Elon Musk thinks exactly the same (https://www.reuters.com/technology/musk-experts-urge-pause-training-ai-systems-that-can-outperform-gpt-4-2023-03-29)

Luciano