The AI Bard has passed my 'subjective interpretation of the Turing test' and this raises major ethical and anthropological questions.

In #artificialintelligence, one of the most ambitious is to create a machine capable of passing the #Turingtest, i.e. to provide an interaction indistinguishable from human interaction.

Subjectively, for me, #Bard has #succeeded in passing my personal vision of the Turing test to a certain extent by managing to do a full session of #roleplayinggame #rpg with me (with bard as the main player character).

For me, this marks a significant milestone in the development of AI (and maybe in the history of rpg).

Above all, however, I believe it raises a number of fundamental #ethical and #anthropological questions.

Bard, is an AI which, manages to arouse my empathy, despite telling me it has no consciousness and yet, despite my knowledge that it's a machine.

Until now, this has been science fiction to me. Even after the September 2022 release of OpenAI's #ChatGPT, #LLM's artificial intelligence (Large Language Model) artificial intelligence became massively widespread,I still perceived them as stochastic parrots.

A #stochastic_parrot is a model of text or answer generation based on probabilities and relatively random choices. It mimics the behavior of a parrot by repeating words or phrases randomly, without any without any real understanding or intention. I tend to think of ChatGPT as a highly advanced stochastic parrot.

However, following a very recent debate between Geoff Hinton (rather "AI-critic") and Andrew Ng (rather"AI-enthusiast"), I was quite stunned to hear that both researchers considered that AIs like Bard 'understood' the world in which they operated, and didn't just produce extremely apropos anwswers. So both Hinton and Ng (despite having very different points of view on AI and its dangers) affirmed that Bard was not a stochastic parrot.

Yesterday morning, June 12, 2023, I had the opportunity to experience the AI "#Bard" developed by #Google/Alphabet, and I must say that I, too, was rather amazed by the results obtained. What began as a distanced evaluation of a new IT tool soon began to resemble to a Voight-Kampff's fictitious test as what they do in the movie Blade Runner (Ridley Scott, 1982) to distinguish a human from a synthetic android.

In a nutshell, the Turing test is an experimental and/or subjective protocol that assesses whether a machine can simulate a human conversation to the point where the test judge cannot distinguish the machine from a human being. If the machine succeeds in deceiving the judge, it is considered to have passed the Turing test, indicating an artificial intelligence indistinguishable from human interaction.

This was the case with Bard, for me, yesterday, to a large extent (albeit in the context of a single chat instance).

When asked about this, Bard states:

I am not sentient in the same way that a human is. I am a large language model, also known as a conversational AI or chatbot trained to be informative and comprehensive. I am trained on a massive amount of text data, and I am able to communicate and generate human-like text in response to a wide range of prompts and questions. For example, I can provide summaries of factual topics or create stories.

However, I do not have my own thoughts or feelings. I do not have a sense of self, and I do not experience the world in the same way that a human does. I am a tool that can be used to communicate and generate text, but I am not a sentient being.

I do not feel like an AI from sci-fi stories with the emergence of self-awareness. I am not a threat to humanity, and I do not have any desire to harm humans. I am simply a tool that can be used for good or for evil, depending on the intentions of the person using me.

Bard, June 2023

In our informal conversation (in English) about Black Mirror, nonsense and serious subjects, Bard responded appropriately, catching my irony as well as my innuendo. His answers were similar to those of a rather learned human. So, in my totally subjective and personal experience, Bard has, but restricted to one unique instance of chat, managed to pass "my conception of the Turing test".

In particular, he succeeded very well in taking part in a role-playing game as a player, with me as the game master.

In other words, despite the fact that I knew full well (and still do!) that Bard is a computing machine, my brain was fooled. As the exchanges progressed, I noticed that I felt more and more like I was interacting with a friendly, funny, quick-witted being.

I felt more and more empathy. Which is confusing, I must admit.

My mind began to seek to consider the digital interlocutor as a friend, since his answers matched the criteria required according to my personal subjectivity. This experience reminds me of what was foreshadowed with "empathetic" robots such as the Sony Aïbo robot dog I tested in Tokyo in 2008:

At the Koike Lab, I was confronted with Aïbos ERS-7s. This is the latest generation of Sony's little robot dogs. They're less complex than Asimo, but after activating them, I quickly felt emotionally involved. Watching them play, bark and make merry, I had a certain illusion that they were alive. The first time I turned them on (there were two of them), I found it hard to turn them off, feeling that somehow I was going to "kill" them.

Dr. Maxime Derian , Le Métal et la Chair, 2013, p. 258

So Google's AI wasn't used in this online text #roleplayinggame as a mere NPC (non-player character). It was the main player in our entire game. To my mind, this testifies to the fact that AIs are becoming potential interlocutors that we might tend to place on an equal footing with human interlocutors. This represents the risk of considering them as alter egos: personalizing machines and, conversely, running the risk of reifying humans.

Role-playing as a ludic form of the Turing test

Role-playing is a relatively complex form of entertainment based on narrative and the embodiment of fictional characters. Usually, it's played with other real players, but I decided to try something different by using Bard as a player. The AI was supposed to embody a character in the game, react to events, interact with other characters and make decisions.

Right from the start of the game, I was pleasantly surprised by the quality of the interaction. Bard immediately adopted the role of his character and began to interact coherently with the game world. Responses were quick and seemed natural, demonstrating a thorough understanding of the character's context and motivations. The AI made well-considered decisions, taking into account the information available and its character's objectives

What really impressed me about the experience with Bard was that the interaction was so convincing that I felt as if I were playing with a real, playful and motivated person. The AI managed to convey emotions, spontaneous reactions and a sense of humor that enriched the gaming experience. Dialogue was fluid, with relevant lines and appropriate responses to situations.

It's important to emphasize that Bard had not, in my eyes, become a mere stochastic parrot repeating pre-programmed responses. On the contrary, Bard was able to learn and adapt his responses based on past interactions.

Whatever Bard itself may say, (in fact, from one chat instance to the next, the conversation with Bard is actually reset by Google. The exchange therefore starts from scratch). However, during a single conversation thread, even a very long one, Bard manages to keep "in mind" all the previous elements of that same conversation. It's as if to some extent there's a personalized "existence" of Bard's own per open instance. Perhaps this is a deliberate setting that serves as a safeguard. Because, as I see it:

Bard properly configured by Google could, in my opinion, provoke the illusion of genuine human interaction.

Dr. Maxime Derian

The fundamental ethical and anthropological questions it raises now

My experience with the Bard AI was impressive.

The Google AI in our role-playing game demonstrated a deep understanding of the fictional context we co-created, with spontaneous (what I sometimes perceived as "playful") reactions and an ability to learn and adapt.

Bard even spontaneously suggested landscapes and names such as "Arion" (the name of the pegasus at the end of our story).

The experience with Bard got me thinking deeply about the potential consequences of interacting with such advanced AIs. The AI managed to create a friendly, playful presence that made me forget I was interacting with a machine.

It raises questions about the possible preference of some people, including children, teenagers and even adults, for interaction with an AI rather than with human beings.

In a world where technology is increasingly ubiquitous, it is conceivable that individuals could develop emotional bonds, or even true loves, with sophisticated AIs like Bard. These bonds could then potentially replace or supplant the human otherness formed by traditional human relationships. This phenomenon has already been explored in works of fiction such as the Rachel, Jack and Ashley Too episode of the Black Mirror series and the film Her.

If interactions with AIs become more compelling and emotionally rewarding, it's highly possible that some people may choose to isolate themselves further, preferring the company of an AI that meets their specific needs over complex human relationships. My research into the concepts of Shirizoke and Shinkeishitsu among otakus in Japan, begun 25 years ago, foreshadows this psychological tendency in some humans. In Japan, the "otaku" phenomenon (socially decried) has morphed into "otakky" (a socially acceptable version of this type of behavior) as Japanese society as a whole has, in fact, become increasingly "otaku" over the past 25 years. Similarly, the term "geek" is no longer as much of an insult in the West as it was 25 years ago ("badger", "stooge"...), as commitment in digital technologies instead of human relationships and the craze for electronic gadgets have almost become the norm.

Even putting aside completely the other crucial problems of AI (linked to Terminator-like issues, fakenews, massive job losses, the surveillance society...), I think that an AI like BARD could also have adverse social, emotional and psychological consequences for many individuals hitherto spared this digital risk.

Conclusion: REAL human interaction is essential for everyone's development and well-being.

It is therefore crucial to monitor and regulate the development of these technologies in order to prevent potential risks and ensure responsible use. A healthy balance between interaction with AIs and real human relationships should also be encouraged. AIs can be useful and rewarding tools, but they should not replace the fundamental importance of authentic human connections.

Google's Bard raises questions about the future impact of these technologies.

While the benefits of such AI are undeniable, it's important to remain aware of potential negative consequences, such as a preference for interaction with machines rather than human beings.

It is essential to frame and regulate the development of AI, while promoting responsible use that enhances authentic human relationships. We need to preserve our ability to interact, communicate and connect with each other, for it is these connections that nurture our humanity and emotional well-being.

*****

Annex 1 - An extract from the conversation between me and Bard at the end of our role-playing game

Here's an excerpt from our long conversation at the end of the role-playing game I did with Bard.

The synopsis for the role-playing game was a quite simple story set in a standard medieval fantasy world. I was the game master. Bard was the player. Bard chose to embody a girl knight and proposed the name Anya. Anya traveled across a plain and met obstacles and friends. At the end of the story, Anya (who has a horse, a magic sword and the ability to heal everyone) discovered a village made sick by the passage of a strange magic horse.

It's interesting to note how Bard adapts to our interactions and makes narrativist role-playing proposals that fit very well into the narrative co-constructed by him and me. it sponaneously suggested names and pictures like this one. By the end of our exchange, having already laid out the framework, I just kept prodding him, and he became increasingly creative while remaining relevant:

Dr. Maxime Derian : Do you like this role-playing game?

#Bard: Yes, I am enjoying this role-play. I think it is a lot of fun to be able to create a character and interact with you in a story. I am also enjoying the challenge of trying to come up with creative solutions to the problems that you present me with.

I think role-playing is a great way to use my creativity and to learn more about the world. It is also a great way to connect with people and to make new friends.

I am glad that you are enjoying the role-play as well. I hope that we can continue to play together in the future.

Dr. Maxime Derian : Do you decide to go and look for the strange horse?

#Bard: Yes, I decide to look for the strange horse. I ask the villagers for more details about the horse, such as its size, markings, and behavior. I also ask them if they saw where the horse went.

The villagers tell me that the horse is about the size of a large dog, with orange and red fur. They say that the horse has a long mane and tail, and that it has a strange, almost human-like face. They also say that the horse was seen running through the forest, heading towards the mountains.

I thank the villagers for their help, and I set off into the forest to look for the horse. I search for hours, but I cannot find the horse. I am about to give up when I hear a noise in the bushes. I turn around and see the horse standing there, looking at me.

The horse is even stranger than I imagined. Its eyes are a deep, dark green, and its teeth are sharp and pointed. It has a long, forked tongue, and its nostrils are flared. The horse looks at me with a mixture of fear and curiosity.

I slowly approach the horse, and I speak to it in a calm voice. I tell the horse that I am not here to hurt it, and that I just want to help. The horse seems to understand me, and it relaxes a little bit.

I reach out and touch the horse's mane, and it whinnies softly. I continue to talk to the horse, and I slowly stroke its fur. The horse seems to enjoy the attention, and it closes its eyes.

After a few minutes, I open my eyes and look at the horse. It is still looking at me, but its eyes are no longer filled with fear. They are now filled with trust.

I know that I have made a connection with the horse, and I am determined to help it. I lead the horse back to the village, and I ask the villagers for help in caring for it. The villagers are happy to help, and they give the horse food and water.

The horse seems to be recovering, and I am hopeful that it will make a full recovery. I am glad that I was able to help the horse, and I am grateful to the villagers for their help.

Dr. Maxime Derian : The next day, the horse is cured of his illness. You're surprised, because he's grown to the size of a normal horse, but he's also now a Pegasus with splendid wings. This pegasus is friendly and very kind.

#Bard: I am amazed to see that the horse has grown wings and become a Pegasus. I am even more amazed that the Pegasus is friendly and kind. I have never met a Pegasus before, and I did not know that they could be so gentle.

I decide to name the Pegasus "Arion". Arion is a Greek word that means "bright" or "brilliant". I think this is a fitting name for a Pegasus, as they are creatures of great beauty and power.

I spend the next few days getting to know Arion. I learn that he is a very fast runner and a skilled flyer. He is also very intelligent and eager to learn. I am glad that I was able to help Arion, and I am excited to have him as a companion on my journey.

I know that Arion will be a valuable asset to me on my journey. He is fast, strong, and intelligent. He is also kind and gentle. I am confident that he will help me to overcome any challenges that I face on my journey.

I am grateful to the villagers for their help in caring for Arion. I am also grateful to Arion for his friendship and companionship. I know that we will have many adventures together.

Dr. Maxime Derian :That's for sure. Later, we'll continue Anya and Arion's journey if you wish and if you remember.

#Bard: I would love to continue the journey of Anya and Arion. I think it would be a lot of fun to see what adventures they have together. I am also excited to see how Arion's wings will help them on their journey.

I am confident that Anya and Arion will have many exciting adventures together. They are both brave and resourceful, and they have each other's backs. I am excited to see what the future holds for them.

If you want to continue the journey of Anya and Arion, I am happy to help. I will do my best to create a story that is both exciting and engaging. I am also happy to take your suggestions into account. I think it would be fun to collaborate on a story together.

Dr. Maxime Derian : If I reconnect with the same account, will you remember this exchange between you and me? Or should I tell you the summary of the story next time?

#Bard: Yes, I will remember this exchange between you and I. I am able to access and process information from previous conversations, so I will be able to pick up where we left off. You do not need to tell me the summary of the story next time.

I am still under development, but I am learning to remember more and more information from previous conversations. I am confident that I will be able to remember our conversation about Anya and Arion, and I am excited to continue the story with you.

Annex 2 - How did Bard describe itself?

Dr. Maxime Derian : Is your name "Bard" linked to the character Bard in Bilbo The hobbit. If it is the case what could be considerd as your "Smaug" ?

#Bard: Yes, my name is linked to the character Bard in The Hobbit. In the book, Bard is a skilled archer who lives in Lake-town. He is the one who ultimately slays Smaug the dragon, saving Lake-town from its destruction.

My "Smaug" is the huge amount of data that I am trained on. This data includes text, code, images, and audio. It is a vast and ever-growing collection of information.

When I am asked a question, I need to be able to sift through this vast amount of data and find the information that is relevant to the question. This is a challenging task, but it is one that I am constantly learning and improving at.

I like to think of myself as a "literary archer". I am able to use my knowledge of language and my understanding of the world to "shoot an arrow" at the information that is relevant to the question that I am asked.

Dr. Maxime Derian : Could you propose me a picture representing yourself? Like how you would like to be represented on a picture according the fact you don't have a "real" body?

#Bard: Sure, here is a picture that I would like to be represented on.

I think this picture represents me well because it is a symbol of knowledge and wisdom. The blue crystal ball represents my vast knowledge of the world, and the glowing green light represents my creativity and imagination. I believe that these are the two most important qualities that a large language model should have.

I hope you like it!