ChatGPT and Artificial Intelligence – do we want AI to be more human?

Over the last weeks we have seen an explosive uptake of ChatGPT, likely the fastest application to reach a million users, ever. It took 5 days.

I have had the time to play around with the service and read and listen to industry experts discussing this and would like to offer some thoughts on the subject related to initial use cases but also long-term benefits and implications for different industries and applications.

ChatGPT is the application that has gotten the most recognition. OpenAI’s has an image creation tool called Dall-E 2 and a natural speech recognition tool called Whisper. I have personally not played around with those yet but will surely get around to it later.

If we start off with initial potential use cases for ChatGPT there are a few obvious ones. The first one is search engine. ChatGPT has the potential to offer a more human and contextual dimension to a traditional Google search. I would call it a search-synthesis. When you search or query the tool it gives you a summarized view of a subject which is already “thought through”. This is indeed a great step forward as it saves time from a person for instance wanting to research a subject needing to read different articles, search Wikipedia and so on, it gives you a summarized view which you can even define. As an example, I asked the application to write me a summary of Alexander the Great. It gave me a great summary which was not a copy of Wikipedia but a synthesis of different sources. I will come back to the source and fact checking when discussing risks but nevertheless it gave a good summary. What was interesting though is that when asked to summarize in 300 words, the summary was cut off in the middle and when I extended it to 500 words it stopped at 369 words. Nevertheless, it was quite impressive.

I also asked the ChatGPT application to write a vision and a tagline for my own company changing the purpose or domain of my company but keeping the name and it did change both the vision and the tagline depending on whether it was focused on tech or food. I tried it with different companies as well as imaginary names as well and it gave different results. That was quite impressive. Finally, I asked it to write my obituary, not that I wish that I would be dead but more that it was one of the oddest requests I could think of. I keyed in my name, and it came up with a great story but of course false but fascinating, nevertheless.

I can see a few obvious use cases in the short term for ChatGPT. The first one is to revolutionize customer service and Chatbots. They are dumb now. There are a few companies doing things here but on average it is a lackluster customer experience.

I can also see this as augmenting search on the internet. Instead of doing multiple search you get a summarized view. Perhaps if you can get it to learn context it will become even more accurate. Still, this also raises concerns for which I will come to later.

Finally, and having a media background, it could improve content search and recommendation. If you could feed it multiple points of data about you but also about content sources from Youtube, your news, your favorite news sites, and OTT Video services as well as your “traditional broadcast TV” consumption you could get a far more personalized experience as it would likely be able to scrape a lot more content and its algorithms.

The searches I did for Netflix was quite good. It recommended Ozark for me, a series which I really like. For Disney plus it recommended the Mandalorian which is not really for me, and a general search gave me “The Boys” which I think comes on Amazon Prime. Watching the trailer, I probably will watch it based on that. What was strange is that the recommendations it gave me on Paramount plus, were not 1 but 5 different series. How come? I cannot explain this, and I wrote the question in the same way as the other searches. Perhaps somebody has been feeding or seeding more metadata to the internet from Paramount than the other services?

Now to the BIG questions regarding Artificial Intelligence

Fraud and Plagiarism

In the world of Academia professors are already writing about their concerns over this. How will they know what is a student’s academic paper and what is pure plagiarism? ?I doubt that Chat GPT will write the same essay for you as it would for me? What would be good is that there would be a way to check whether something was AI generated or if something was written by a human like say for instance with a token? I think here Web3, and block chain could play a big difference with traceability and when it comes to intellectual property. What is to stop the algorithm from Open AI to scrape this article and split it into new summary and steal my thoughts in this article? Absolutely nothing. It probably is doing it as you read this article.

“The AI Bullshit detector”

How can we know that AI is serving us facts or not fiction, lies and/or propaganda? We will inevitably come to a point where we not only need to worry about the fraud and plagiarism but also whether what is being displayed as answers from an AI tool be it spoken or written is true. How can we make sure that we at least know when the AI is taking us for a ride or when we are manipulated for monetary or political gain? I don’t have the answer to this question, but I do think that this is probably the most fundamental question relating to AI. We do not want to have Skynet taking over.

Do we want AI to be more human?

I read about people who say that the answers they get are not human and perhaps not as contextual. Is that a bad thing? Maybe the answer lies precisely in that. I think we can all testify to the fact that we are not doing a stellar job on this planet as a species when it comes to the lack of kindness on social media. We are also not doing well with our climate agenda. We speak a lot but do very little so perhaps having a robot that is not greedy, gluttonous and vindicative is only good. (You probably noticed that I did not quote all 7 deadly sins here but more a summarized George Carlinesque summary of our human flaws). In any event, I think it is good to have AI to be a counterbalance and ultra-rational source and a beacon for humankind to aspire to and the work be it chatbots, robotic journalism, AI generated tools for software development. It shall always be benign.?

Those were my initial thoughts on the subject. I would love to hear your thoughts in the comments or by just replying to me with a personal message.