ChatGPT: From the Turing Test to the Baricco Test.
ChatGPT and other AI-based text generation systems have achieved the unthinkable: Wikipedia is now considered a reliable source of information. Now it is better to be wrong with the other sapiens than to get trapped in the mysterious algorithmic hallucinations of OpenAI… What does the collective machine created by Jimmy Wales say about the Turing Test?
The Turing test, originally called the imitation game by Alan Turing in 1950, is a test of a machine’s ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human. 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.
The Turing Test went down in history: today, in June 2023, we can spend a whole weekend locked up with ChatGPT talking about the pre-Socratic philosophers and, if it weren’t for the media hype, we would hardly realize that we were chatting with a ‘smart’ machine. Or something like that.
But Alan Turing was not the only one who came up with the idea of inventing a method to identify non-human intelligences (Digression: pay attention to this concept, ‘non-human intelligences’, because I wouldn’t be surprised if we soon start using it instead of ‘artificial intelligences’. It is a much more appropriate concept to name the new forms of intelligence with which we must learn to live with, in the same way that in recent years we have become aware of the existence of other intelligences, from plants to cephalopods. End of digression).
The Voight-Kampff Test
The famous Voight-Kampff Test that appears in Blade Runner (Riddley Scott, 1982) has already entered the history of science fiction. Any human or replicant remembers that excited eyeball on a little screen:
Rachael: Do you mind if I smoke?
Deckard: It won’t affect the test. All right, I’m going to ask you a series of questions. Just relax and answer them as simply as you can. — It’s your birthday. Someone gives you a calfskin wallet.
Rachael: I wouldn’t accept it. Also, I’d report the person who gave it to me to the police.
Deckard: You’ve got a little boy. He shows you his butterfly collection plus the killing jar.
Rachael: I’d take him to the doctor.
Deckard: You’re watching television. Suddenly you realize there’s a wasp crawling on your arm.
Rachael: I’d kill it.
Deckard: You’re reading a magazine. You come across a fullpage nude photo of a girl.
Rachael: Is this testing whether I’m a replicant or a lesbian, Mr. Deckard?
Beware of replicants!
Baricco Test
The Game. A Digital Turning Point (McSweeney, 2020) by Alessandro Baricco can be read as the continuation of one of his most outstanding and acute essays: The Barbarians: An Essay On the Mutation of Culture (Rizzoli, 2014). In The Game Baricco drops very strong ideas about the tensions and conflicts generated by digital technologies / cultures:
“Look at the map upside down.”
“New humans are not the product of smartphones, they are their inventors.”
“If the digital revolution scares you, invert the sequence and ask yourself what we were escaping from when we opened the door to a revolution of this kind.”
“Get used to thinking of the digital world as an effect, not a cause.”
But Baricco goes further. According to the most POPular Italian thinker of his generation, the logic of the video game has already left its mark on the life of Homo sapiens. But Baricco is not talking about the much celebrated ‘gamification’. He really goes further:
“I’ll be a little brutal here: for historical — possibly Darwinian — reasons, from a certain point in time (since the iPhone, I would say if I had to hazard a date), there was no chance of anything surviving if it didn’t share the same DNA as video games. I’ll even compile a useful checklist of the required genetic characteristics of the species needed to survive.
And what are those “genetic characteristics”?
“An attractive design that gives sensory satisfaction;
A structure that can be reduced to the repetition of an elementary problem/solution pattern;
A minimum time lapse between the problem and its solution;
A progressive escalation in the difficulty of the game;
No need and no use for being immobile;
A learning curve provided by the game itself, not by studying abstract instructions;
Immediate accessibility, without any explanation;
The reassuring presence of a point system after a certain number of steps.”
Baricco concludes:
“That’s about all I can think of for now, but I have some important news for you: if you are engaged in an activity that doesn’t have at least half of these features, you are doing something that has been dead in the water for some time.
You are allowed to be a little unsettled by this news.”
We can invert the Baricco Test: “if you are engaged in an activity that has at least half of these features, you are doing something that is really alive”.
Now, let’s talk about ChatGPT.
Why are we so excited about ChatGPT?
I imagine that you already understand where the shots come from. The explosive success of ChatGPT can be explained by the following fact: the technology that OpenAI unwittingly released without considering the consequences at the end of 2022 complies with most of the principles of the Baricco Test.
- ChatGPT has a attractive and simple design that does not need an instruction manual.
- ChatGPT has a structure that refers to the elementary schema problem/solution repeated several times. We ask, the machine answers, we ask again, etc. When the machine interrogates us (don’t worry, that day will come) we will know that we are the subject of an inverse Turing Test.
- In ChatGPT there is minimum time lapse between any problem and its solution.
- If we wish, we can progressively increase the difficulty of the questions we ask. It is a ‘chat’, don’t forget: we should never keep the first answer, we have to go for more.
- For now the ChatGPT is a ‘fixed’ experience. In other words, for now it does not respect the principle of “no need and no use for being immobile”. For now.
- To use ChatGPT we do not have to read a manual or follow instructions. Learning “provided by the game itself” and not by “studying abstract instructions.” ChatGPT is a great example of learning by playing, the opposite of following IKEA’s instructions when assembling a piece of furniture
- ChatGPT assures us “immediate accessibility, without any explanation”. It moves in the dimension of the algorithmic time -or, if you prefer, it lives in the temporal scale of the “infinitesimal time capitalism” described by Agustín Fernandez Mallo in La forma de la multitud (Galaxia Gutenberg, 2023)-, a dimension that is very difficult for Homo sapiens to grasp. AIs live on another time scale both because of their ability to process information and manage millions of data, hence ensuring instant enjoyment.
- While they still don’t incorporate a points system (Baricco talks about the “the reassuring presence of a point system after a certain number of steps”), it’s not hard to imagine possible reward mechanisms within smart chatbots.
Game over?
As I said a couple of months ago at a conference in Santiago de Compostela, everything we write or say about AI is already old. Such an accelerated phenomenon of mass diffusion, use and experimentation of a technology has never been seen before. Not when software bursted in the early 1990s, not with the spread of the World Wide Web after 1995 or the explosion of blogs or social media at the beginning of the new millennium. The closest thing is perhaps the big bang of Pokemon in the summer of 2016: it was the first massive and global experience of Augmented Reality usage. But this time we are facing a truly disruptive phenomenon
In the last six months, millions of people all over the planet began to generate texts, ask questions or create images. It is an easy, nice and cheap technology. In other words, millions of Homo sapiens discovered the joy of playing with artificial intelligence. Welcome to The Game.
The same is happening at the level of scientific publications: hundreds of papers and books on ChatGPT are being published every week. Something similar is happening with interfaces: ChatGPT and similar systems are being integrated into all kinds of technologies or services. And we already now that when a new technological actor (in this case AI) enters an interface, the entire interface changes: relationships are modified, processes are transformed, and some actors run the risk of being marginalized or outright cancelled. By far surpassing the Baricco Test, ChatGPT and other intelligent systems fulfill all the conditions to become the killer application of the decade.
And in an unexpected script twist, in this thriller it only remains to figure out who the victims will be.
Bonus tracks
- “Artificial Intelligence: Half Wanting, Half Fearing” by Carlos A. Scolari in CCCB.
