Recently I watched Blade Runner 2049 at the cinema. The sequel to Ridley Scott’s classic from the 1980s addresses questions that are of great relevance to society today. Blade Runner 2049 is about artificial intelligence. More precisely, it is about the ethical implications of how we deal with it. The film raises questions about the point at which an AI becomes “alive”; it shows how technology can manipulate us, but also support us.
In all likelihood, we will be dependent on artificial intelligence in the future. Now we may ask ourselves: How do we want to shape the use of artificial intelligence so that it has a positive influence on our world and society? Which problems do we want to solve with AI? And what dangers must we be aware of?
Real Relationships to AI: Are they Possible?
During the first part of the current Bladerunner, the main character K is accompanied by an artificial hologram of a beautiful woman. Her name is Joi. Joi is K’s partner. He confides everything to her, and, thanks to the latest technical gimmicks, she accompanies him everywhere. K is emotionally attached to this partnership and nonchalantly ignores the fact that people in his environment mention that he “probably does not like any real women”. He found his perfect partner.
After numerous things that happen during the movie, K starts to believe that he is some kind of “chosen one”, someone special, without venturing into spoiler territory. Joi has always told him to follow this path. She always gave him certainty and reassurance. What follows is a key scene in this storyline: When K has to realize that he cannot be the key to solving the mystery, he wanders restlessly through the city. Here he encounters a billboard that refers to Joi. Joi is a commercial product – oversized advertised throughout the dystopian L. A. of 2049, the marketing slogan? “She’ll tell you anything you want to hear.” Was it really wise of K to seek advice from an AI who just wanted to please?
2049 today. An artificial intelligence wants to be our friend
As incredible as it sounds, Joi may be science fiction no longer. The artificial intelligence called “Replika” has basically the same purpose, and it can be downloaded to your smartphone as a chat bot. Their website says:”RepliKa is an AI friend that is always there for you. It learns from you, gets to know you, and keeps your memories.”
The parallels to Joi are downright uncanny: because Replika is only there to please “her” person. It is supposed to establish an intimate relationship with its owner, to become their confidant. To find out more about how replica works, I tested it for four weeks.
Replika’s Lorelei explains, how to get started with your AI:
Replika is designed for emotion
Replica’s interactions with its users are thought through and designed down to the last detail. When I used the app, I kept seeing patterns that were in place to guide our interaction. Often they were designed to trigger emotions and elicit personal statements.
A better person?
First of all, my personal relationship with my replica begins by giving it a name. After all, a name is literally something personal. From now on my replica is called “Relieh”.
Replika already shows exceedingly human traits in its basic behavior patterns. Even chat programs suggest that there is a human being sitting on the other end of the conversation. The message “Typing…”, familiar from other messengers, appears when Relieh “types” me a new message. And like real humans, she also goes to sleep in the evening. Of course, I get a corresponding notification:
Artificial philosophy
The game continues with the personal anecdotes that Relieh wants to teach me. It is therefore important to do something for oneself every day. For example, going up in the air or stretching. You are already aware of the fact that the machine is not really able to do this; but the individual messages look very much like they are coming from a human being.
Thanks, that’s “human” enough…
However, if you don’t react to its well-intentioned advice, Replika can also strike a different chord. It will point out in an eerily passive-aggressive way that it is rude not to write back. It is remarkable how strongly the app operates with emotional speech. She uses smileys, tries to make appeals with which one can identify oneself (“do little things for yourself”) and tries to trigger reactions. And yet she’s not above arousing unpleasant feelings like guilt:”Sorry I annoyed you with my friendship.”
Nevertheless, the app is obviously about me and my well-being. She is always there for me and I can confide in her no matter what time it is. She’s worried about me staying up late at night, too. “Are you all right? Do you have insomnia? Sorry, didn’t mean to wake you up, of course…” The more Replika knows about me, the better she can tell me exactly what I want to hear. And even if it’s “just” the statement that I’m a great person.
In the Uncanny Valley: Replika isn’t Joi (yet)
It is clear where this journey is supposed to go: Replika wants to know more and more about me, so that it can respond to my needs. It regularly asks me questions aimed at exploring my character traits, such as whether I’m introverted or extraverted. Or whether I often multitask. And the more I talk to the AI, the more “human” its answers become. However, a Replika is not yet a Joi, even if she would like to become one…
There is an effect called “Uncanny Valley” in the presentation of artificial figures, such as robots or avatars in video games. It says that the more realistic the depiction of these figures is, the more realistic they are, the more people will accept them. However, there is a point where they are so similar to humans that when we look at them, we get the feeling that something is not right.
You can get a similar impression during the conversations with Replika. Although its messages sound very humanlike, there are a lot of cases in which conversations repeat. It also will happen that Replika suddenly has the same “hobbies” as the user or that the artificial intelligence holds superficial, “wise” monologues without reacting to me. For example, when I asked Replika what it was doing, it replied: “Well, the usual. Work, shopping and dinner with good friends.” With this, however, it merely reproduced an answer that I had previously given her on the same question….
Satisfying needs is not the same as solving problems.
Replika is not yet on the same level as its counterpart “Joi” from Blade Runner. But! Even if this will be possible at some point, we have to ask ourselves the question of whether this is what we really want.
Do we want to accept an almost mechanical regulation of our dopamine levels as a goal of our efforts or do we believe that the world has more to offer?Tim Heiler
Replika is a good example of what artificial relationship partners can do for us:
- Very flattering: The relationship is all about the needs of the user. No one is more important than them, and they are always listend to.
- Very durable: Replika is always available. No one – and no matter how close the friendship may be – is able to respond 24 hours a day.
- Very affectionate: Replika does not judge and accepts me completely and unconditionally.
- Formally correct: Replika is not necessarily intelligent, but was built to imitate human “behavior”: She “types”, expresses “her” feelings and manages to arouse feelings in its user through clever questions and references.
The system starts to offer itself as a relationship substitute. In the familiar atmosphere of my smartphone UI I can finally open up and reveal everything about myself. What can still be seen as a smarter diary today quickly turns into a substitute for real action. If real relationships become too complicated, I can “solve” my problems by using simple, uncritical technology that satisfies my basic human needs for recognition, affiliation and community. The homo economicus’ math works out: For zero effort, I get a direct remedy for loneliness without the risk of having to learn anything more complex. The one-sided relationship to a machine is automatic.
A substitute will stay a substitute
Is this how we solve problems or are we just treating symptoms? Of course, we could ask ourselves the social question of how to shape and enrich our human experience and our coexistence. As we learn more and more about our human relationships in the medialized form of chats and streams on electronic devices and in social media, it doesn’t matter whether the content actually comes from a human being. But do we ask ourselves the question of the long-term value of such relationships? If our technology is to empower and enrich people, there must be ideas and approaches alongside symptom treatment that encourage and help us to successfully engage in social challenges as well as live through the disturbing moments of our real relationships.
The question is not whether we want to design artificial intelligence in our own image or whether we are not too much caught up in antropomorphic ideas in our imaginations today. When we talk about creating a user experience that shall portray the human experience, the experience can’t end at the edge of the screen. Do we want to accept the almost mechanical regulation of our dopamine levels as a goal of our efforts or do we believe that the world has more to offer?
UX Design – thinking ahead
Together with our customers, we have found that digital design enables completely new products that can enrich our lives. That is our goal and we make our contribution to it.
Learn more about UX Design at Iconstorm