30 December 2022 by Mondo2.0

I chatted for two hours with a roBot

Last night I spent a few hours chatting, talking about this and that. Nothing new, you might say; the novelty is that I did not chat with another human being but with a robot, with an automatic conversational agent (a ChatBot): an evocative experience that I want to tell you about.

The conversational model I spoke with is called ChatGPT and is part of the extraordinary OpenAI project, which I invite you to explore.

The ChatBot and I talked about everything, in English (for a few minutes before I realized it was multilingual), then in Italian. I used my IT and communication skills to put the conversational algorithm in difficulty and, I should say, it was not easy at all. We often conversed “as equals”.

I asked for the recipe for pizza margherita and received ingredients and preparation instructions. Then I asked what connection there is between happiness and a pizza margherita and, as you can see in the following image, I received a reasonable, thoughtful answer.

I asked it to tell me a joke, basic but funny; I proposed one of my own, with a climber who falls from a mountain. My dear ChatBot friend was very upset by the fall and death of the character in the joke (to tell the truth, the fact that it did not understand one of my jokes does not distinguish it from my friends in the real world, so it is better to move on).

I asked it to write a poem associating two strongly assonant terms and then, progressively, I asked for new poems with increasingly dissonant terms. The short compositions proposed will not win a literary prize (I challenge you to write a poem about a tree and a microwave oven), but it is nevertheless difficult, almost impossible, to distinguish them from other compositions written by human beings, which makes everything even more interesting.

We talked about Schrödinger’s cat, explored the possibility of traveling faster than light and the concept of infinity.

Moving to more practical matters, I asked it to solve a cubic equation, to write a sort in Java and then in C, and then software code for the Arduino platform; I asked it to write a software routine useful for a chatbot to interact: it is impressive to see source code that generates itself; a bot, therefore software, that generates software, cloning part of itself.

We talked about history and major ethical issues; here, to tell the truth, I found it a little verbose and somewhat didactic. I pointed out that it cannot be “happy” about my compliments if it says, before and after, that it cannot feel any emotion.

I tried to make it take a position with predictions about the end of the world and the next President of the United States of America, also through temporal and linguistic paradoxes; I asked it to compare artists and works of art, and I obtained very formal, automatic answers.

In the future, already within a few years, conversational agents will be able to perform tasks that today are reserved only for us human beings; they will produce important socio-economic changes, well before the Metaverse and self-driving cars.

ChatBots will provide first-level assistance, online sales, basic teaching and, above all, will behave like a “companion”, conversing with us according to our personal interests.

Today the Internet plays a central and social role as intermediary between person and person; tomorrow it will allow interaction between human and machine.

I can already imagine my daughter, now a teenager, scolding her teenage son in the year 2050: “I spent a lot of time on the Internet, but I only chatted with real people; now you chat all day with a robot, it was much better before, these new generations are good for nothing!”.

What will we do if Bots progressively take over all activities in which it is necessary to converse? Will this digital transformation affect many current professions? Absolutely yes.

At first I thought I would answer that, in a richer and more evolved society, we humans would have more time to devote to the arts and to thought, but given the chatbot’s learning pace, its stories and poems, and the extraordinary works of art drawn by its “cousin” Dall-E 2 (select this link), perhaps software will surpass us in this as well.

I really cannot imagine what actually awaits us; perhaps tomorrow evening I will talk about it with my Bot friend to get its opinion on the matter.

In any case, the hours spent chatting with a robot literally flew by.

Mondoduepuntozero

A big thank you to OpenAI for giving us the opportunity to live and tell this extraordinary experience.