ChatGPT: Remembering Turing with some Fun Trials

The arrival of ChatGPT, OpenAI's new chatbot feels like a pivotal moment in our lives. If you are haven't already checked it out, I recommend that you do. I give a few examples below to give you sense of how good it is.

Alan Turing, the mathematician devised a test to see if a machine has achieved human intelligence. 

From Wikipedia:

Turing proposed that a human evaluator would judge natural language conversations between a human and a machine designed to generate human-like responses. The evaluator would be aware that one of the two partners in conversation was a machine, and all participants would be separated from one another. The conversation would be limited to a text-only channel, such as a computer keyboard and screen, so the result would not depend on the machine's ability to render words as speech. If the evaluator could not reliably tell the machine from the human, the machine would be said to have passed the test. The test results would not depend on the machine's ability to give correct answers to questions, only on how closely its answers resembled those a human would give.

By this definition,  ChatGPT seems to have achieved human intelligence. At least, it is much closer to human intelligence than anything that I have seen so far. That makes me wonder if Turing would have been proud or concerned at how good it is.

It writes poems. On serious topics.

And frivolous ones!

I tried to use it a bit for self promotion.  It obliged!

It knows when it has produced something that is red hot. It censors itself!

I asked it to write about my friend's car. It doesn't know anything about it, but that didn't stop it from making something up. A quality I have seen in many people.

It can write a scene. See how good the dialog flow is and how much it seems to "know" about Einstein and Trump.

It knows Shakespeare's style, but does not undersand a phrase from Julius Caeser. It makes up statistics to compensate ('Beggers usually die during the day!'). Again, a human trait!

It is aware of investor sentiments on market crash and even takes the trouble to cheer people up!

It recognizes poetry in the text promt, even if you don't designate it so!

You can use it cheat in crosswords. But it seems to have trouble counting, just like some humans. ("Pity" is a five letter word?)

Finally, it seems self aware. Scarily so! It can keep track of the previous question for context. 

I haven't seen anything this good among chatbots. The only factor that suggests that it's not human is how quickly it writes all this.

We have seen new developments in AI generating images from text prompts. Now it can generate good quality text too!  It's exciting (and troubling) to imagine all ways this can be put to use! 

What do you think?

Turing Image Credit: Wikipedia.