Why Be So Ambitious About a Chatbot, Really? Research Time, 21/06/2016

You know, even though I spent yesterday’s post talking about the global mass fragmentation of online space, most of that actual post Brian Honigman published last week was about chatbots.

Now, chatbots are weirdly fascinating creatures. I mean this in a strictly intellectual sense. Because every time I encounter a chatbot in my actual life, they irritate the hell out of me. And this is a very important point for my argument that you should remember as I go through a more complex reaction to Brian’s ideas.

Hal, I asked you to open the pod bay doors. Why are you
asking me to do your survey about my satisfaction with
Discovery Space Systems before you'll let me back on
board the ship?

In a broad sense, artificial intelligence has a profound hold on the Western popular imagination. This is true for two worlds that I’ve worked in over the last decade – science fiction and philosophy.

Philosophically, I’ve studied cognitive science and the theory of mind in my younger days. Contemporary philosophy of mind developed from the community of scientists and thinkers that were part of the cybernetics community in the mid-20th century. 

This was the community whose ultimate goal was the creation of artificial intelligence. In its original sense, this was the replication of a human-style mind in a machine. Among the philosophical community (at least in academia), artificial intelligence is still largely focussed on this goal.

It’s the same goal and ideal that animates so much of science-fiction, from the first imaginings of mechanical men, through the creation of the term ‘robot,’ the innovations of Golden Age writers like Asimov, the more nuanced explorations of Star Trek: The Next Generation, Battlestar: Galactica, and so many other stories across so many other media.

It’s a focus of some of my own creative work, like the development of my Alice character for the film project Here Is a Man, and my plans for future short stories and novellas imagining the far future of humanity. 

Having survived multiple ecological crises and dystopias to live spread among stars hundreds of light years apart. A species of robotic sages who we created drift from world to world, the mirrors of our best natures, to whom we always fall short.

Dream casting for Here Is a Man: June Diane Raphael as
Alice. She plays the same basic sarcasm and sense of
superiority she does so well on Grace and Frankie, but
where her character also has the genuine wisdom,
perceptiveness, and martial skill of the master sage
described in the Zhuangzi.

The human imagination has developed profound imaginaries of our most hopeful, optimistic alternative futures. And the most visible manifestation of artificial intelligence in our real lives are bloody chatbots.

That says a lot about human nature right there. 

Honigman discusses the potential of chatbots to improve consumer experiences, inquiries, and marketing techniques. Which is genuinely interesting territory – it’s fascinating and sometimes horrifying to develop algorithmic entities that can become creative and innovative thanks to continuing relationships with humans. Intelligence literally manifests through experience with an intelligence.

But our abilities at turning an algorithm heuristic* are still very limited. A chatbot can develop a very detailed and comprehensive power to respond to human needs and inquiries. But only within a very limited conceptual space.

* If I can borrow a phrase from Arthur C Clarke.

AI engineers can’t yet program a chatbot intelligence that can truly learn and adapt with the same flexibility that humans can with our uncommonly plastic, malleable brains. All the parameters of a chatbot’s interactions have to be pre-programmed into it. 

They can learn and improve within those parameters, so the best chatbots are the ones that perform very specific customer service functions. They're the front lines for a business to answer consumer questions in real time about specific product lines and company activities. 

With so much contemporary research showing how rooted
intelligence is in our embodiment and interactions with
our environment, building a truly human-like artificial
intelligence would mean literally building an artificial
human, down to our own neural plasticity and
developmental interaction with our environment.

The only AI we know how to build right now is one in which we can anticipate the boundaries of everything it will be asked to do. Life, taken on the whole, is too complex for anyone to anticipate. To imagine life in advance is beyond human intelligence.

So instead, we build computer programs to interact with people in messenger apps about consumer products. The first crack at help lines and customer complaint logs. A cheaper replacement for human operators.

With a little more flexibility, they can spam us with unsolicited and obviously fake conversations about a company’s products. In just the same way as I get generic emails to “try our progressive new product” from hotmail addresses that contain more numbers than anything like someone’s actual name, alias, or avatar. 

And I’ll pass over those open marketing messages in the same way I pass over those spam emails that slip through my filters and clutter my feed some mornings. Only I'll be more irritated, because I may not be able to get the chatbot to shut up.

Then I’ll go back to writing my story of the robotic sage who wants to save us from ourselves. Or watching an adventure with Geordi, Deanna, and Mr Data. Chatbots might be the future of a lot of marketing channels and consumer-company interaction, but I remain skeptical that they'll ever stop being annoying.