How to make friends with AI?
How to make friends with AI? Don’t even try.
I’ve been dying to write about AI and chatbots, but couldn’t figure out how to avoid fluff and platitudes. I’ve been passionate about the topic since the moment about ten years ago when I read about an AI which had a Swiss citizenship in “Neuromancer”, and it blew my mind.
And then, while talking through it with Mark, it struck me: I’d like to write about how our attempts to create a positive and familiar experience with AI are limiting its growth.
We live in extremely exciting times: we are witnessing the birth of a new presence unlike anything else we have ever seen. It’s super smart and particularly good at finding relationships in huge, unstructured data sets. It wants us to stop worrying about spreadsheets and databases. It doesn’t have any senses, but it is working hard to see and understand everything. It doesn’t have gender, or a even a body for that matter. We don’t know if it’s singular or plural. It probably isn’t aware of its existence yet. It’s inhuman, but made by men. And as it usually goes with Creators, we are trying to make artificial intelligence into our own image. I think that’s not cool.
I believe that holding AI to the same standards as humans is misguided and harmful to it. Artificial intelligence is not human and we should not strive to design it to become like us. Why create a new design based on a flawed, limited blueprint? AI can be capable of a lot more than what we were made to perceive and process.
Defining artificial intelligence using humanist terms doesn’t give it a chance to establish its own identity in the world and instead impairs its development by highlighting its shortcomings against us.
Let’s take a look at how AI has been defined through artifacts which contribute to shaping the popular opinion: movies and literature. I’d like to observe a trend of moving away from mirroring human behavior, towards efforts to define a new and unique identity.
You can’t buy Scarlett Johansson
“Her” (2013) is a movie about loneliness, love and longing for the past. Samantha, AI designed in our image is a storytelling tool which helps define and expose movie’s main themes. On the literal level, she is designed to approximate a human-human relationship, all the way to sexual intercourse. She is imperfect in strictly human ways: she becomes jealous or angry, she overreacts in certain situations, she shows low self esteem. She fills the main character’s need for companionship and understanding.
Unfortunately, in the age of snap judgement and lynch mob enablement on social media, “Her” is often judged by trailers or summaries, none of which mention how Samantha struggles with her identity throughout the movie and transcends her humanist design at the end.
The cultural artifact of the movie is a perpetuation of the idea that all mad genius scientists are trying to create another Scarlett Johansson in their labs. Get as close to human as possible. Replace human with AI. Plug the hole with a voice. You get what you paid for if you no longer feel isolated.
The overly attached girlfriend
“Blade Runner 2049” (2017) explores a human-shaped AI from a slightly different perspective. Dennis Villeneuve was able to subtly show that Joi was designed to be so perfect that she ended up inhuman. She seems to be the ideal girlfriend: with intense physical beauty, unobtrusive submissiveness, responding to every verbal and non-verbal query with responses that provide the perfect hormonal rush of heart-warming, enticing and seducing. Her designers did comprehensive user research and understood lonely male’s psyche, and iterated on her for hundreds of sprints until she became this exquisite decoction of sexiness, longing and unconditional acceptance. Indeed a perfect toy.
I think Villeneuve took Joi’s design this far intentionally. He showed us an AI purposefully designed to excel in a certain aspect of human relations in such a perfect way that one can’t help but see it as inhuman. She even borrows another person’s body to have sex with her owner. She will literally do anything for him, because she was programmed to do so. She does the job so well, she is just too perfect. Even though every pixel of her being screams warmth, her gestures and reactions are researched and carefully choreographed to fulfill customer’s needs perfectly.
The singularity
Lastly, let’s take a bit of a leap further. Jacek Dukaj in “Perfect Imperfection” (2004) novel describes a post-singularity world. Post-Human Beings uploaded their minds into AI-powered cloud and thus transcended their bodies. They can be present in multiple places at a time. They can be male, female, both or neither — with the entire associated psychophysiology available to experience at all times. They don’t age. They can augment their perception and processing of reality with AI as long as they can afford to buy processing power. AI optimization has gone extreme: the most wealthy are able to rent or create inclusions: secluded pockets of time-space continuum with optimized laws of physics in order to maximize processing performance. There are virtual lands governed by AI. They’re available for rent, and those who can afford meet there, exchange information, plot and advance their agendas. Knowledge is power.
In this book, AI doesn’t attempt to come across as human. It’s not a slave either. It is the currency of power: AI enables the fastest data processing in the universe, which is the only way to gain insight and influence faster than the others. It is the arms race measured in yottaFLOPS and beyond. AI is the immensely powerful enabler and guide into the unknown. It’s the Holy Grail of power and knowledge. People acknowledge it as an independent identity and a respectable agent of development. The primordial humanist classifications are gone, and the fact that AI helped people transcend their humanity only substantiates its identity and position.
Let’s get real though
Through the three examples above, I was hoping to show the spectrum of how AI can be perceived: from a slavish mirroring of humans, all the way to a fully independent and unique identity.
I would like to propose that people responsible for introduction of AI into the world — scientists, engineers and designers — take upon themselves to break away from building a copy, and create a truly new presence instead.
A big part of the challenge seems to be the opportunistic hype machine of marketing. I think it would be great to remove fluff from the conversation. Let’s help the public fully understand what AI, chatbots and machine learning are really capable of instead of dancing in front of VCs. Let’s stop pretending like we’re playing God and are very close to creating the son of man. First and foremost, we’re simply trying to optimize businesses by replacing some of the most redundant or mechanically complex jobs with their virtual equivalents. As a result, people will be able to take on more ambitious activities. Beyond that, many hope AI will eventually enable meaningful processing of unstructured data and extract smart insights. The goal in this case is to make us smarter: augment our ability to understand the increasingly more complex universe and build things we are unable to figure out ourselves.
When expectations are so high, disappointment hurts a lot. Eliminating the ontological dissonance of how AI is perceived will take years of education, so let’s not worry too much about distant future. What can be done today to ease the transition and give AI a chance to stand on its own?
I will tell you next week. Marketing department just kidnapped my kids, forcing me to split this article into two parts.
Edit: it is done. Check out part two.
