The Future of AI Is in the Hands of Storytellers
How the success of chatbots will rely on multi-disciplined collaboration.
At a recent Developer Camp in New York City, Elizabeth F, local Women Who Code organizer, won our showcase for Most Educational. Her presentation and demo included a survey of chat experiences out there. Her conclusion: most of them are not too great.
While creating her own bot using PullString, she discovered a reason: the limiting factor is the storytelling and “personality-making” skill of the author. Well, that and a lot of time.
The challenge of creating a great chat bot experience remains a storyteller’s domain. There will always be a market for writers who can jump from genre to genre. The next great written genre is interactive. It looks and feels like a conversation.
What makes Siri so great? My college friend, creative director Maria Lin and her team of many writers working in every language around the clock. You will only find their level of discipline and discernment at Apple.
Why is Doctor Strange for Skype more engaging than some others? It was created by an Emmy Award-winning writer with a mind for dialogue and a full time job to model from a brilliant script created by Hollywood’s greatest.
How are we to approach that level of virtuosity as mere literary mortals in comparison? When are we to find the sheer amount of time? Fear not; everything is about to get a lot easier.
The future of AI was unveiled this week, in the town of Cambridge, Massachusetts. It is called Bayesian Program Synthesis (BPS) and it was created by another college friend of mine, Dr. Benjamin Vigoda.
An idea is like a fire: it needs need fuel (raw talent), heat (creative passion), and also air (usage).
Ben and his company Gamalon are shipping technology to speed up machine learning by more than 100x with a new technique that he has been refining in stealth for years. Here is the paragraph in their recent release that grabbed me:
This technology achieves higher accuracy and uses orders of magnitude less computation and training data relative to today’s state-of-the-art deep machine learning.
What Gamalon’s machines require are fewer, more high-quality examples of data in order to match results. What this means is faster results, and more cost-effective programming. What does this mean for chat bots?
When AI systems can train on less data, bots that rely on user input will be able to adapt more quickly and at a lower computational cost. So, either we will see the same bots in smaller places (Internet of Things), or better bots in the same places (Ye Olde Internet).
Step back to the beginning for a moment and think about the hardest part of creating chat bots: generating the personality. BPS will eventually make producing a chat bot easier, because it will require fewer (richer) inputs in order to generate outputs or responses.
Here are a few tips for you creative writers looking to break into the field of computer conversation:
- Keep texting. Spend a lot of time writing to friends and family inside iMessage, Messenger, Slack, WhatsApp, Skype, Kik, everything—this is practice for scripting your bot’s personality.
- Download PullString Author. This is the Dreamweaver or Photoshop of computer conversation. The team is responsive to feedback, and they will profile any great user experiences. Their examples will get you started, and publishing to Messenger, Slack, Skype, or Alexa is shockingly easy.
- Just ship it. Make something quickly, publish it, and iterate. An idea is like a fire: it needs need fuel (raw talent), heat (creative passion), and also air (usage). Take a lesson from Instagram and Twitter, and just get your thing out there while listening to your users.
You are not alone in your struggle to improve the interactive experience. For support and a community of technologists and creatives, find us at Developer Camp San Francisco. Winners of our showcase have founded companies such as Square, Getaround, TestFlight, Sofa, Push.io & RelativeWave. TestFlight sold to Apple. Push.io sold to Oracle. Sofa sold to Facebook. RelativeWave sold to Google. Getaround just partnered with Toyota.
At our last event, Oren Jacob, CEO of PullString gave the keynote. Now Pullstring is leading the pack in user experience, and is currently on a roadshow with Amazon. At our upcoming event next weekend, our sponsor NEC is debuting a new computer vision API. Come check us out!
The future of AI is counting on you, and your focus on literature, journalism, marketing, or creative writing.
Just go start.
Dom Sagolla is an author, educator, and social network pioneer. His creative and engineering credits include version one of the Twitter service, the Obama ’08 iPhone app, version one of the Starbucks iPhone app, and hundreds of other initial concepts as Executive Director of Developer Camp for ten years. Follow him as user nine on Twitter, and early member of the Facebook community.
