Greetings everyone. I’m Paul Rowe from Vernon Systems. You’ve probably already met Siri, a celebrity in the bot world. Today I’m going to introduce you to some of her friends.
Bots are automated pieces of software that carry out a set of operations with some degree of independence, often with behaviour mimicking a person. They’re becoming increasingly sophisticated and have been adopted for a range of purposes within the museum sector such as sharing random collection items and helping to answer visitors’ questions.
This is one of the first bots I ever met. He was called The Warlock, and he lived on a place called Firetop Mountain, which I believe is somewhere near Springfield. The Warlock was my guide through a Choose Your Adventure book.
You might think I’m joking, but this is a bot, just an analogue paper-based one. This narrative works through a series of questions for which there are pre-written answers. You can pick an option and then get directly to an appropriate snippet of further content. Hot tip: Beware of skeletons.
… and if we jump forward to the present have we have the Auckland Art Gallery chatbot. It can free up staff time by answering simple frequently reoccurring questions.
It knows about what’s on and opening hours and it’s also connected into the gallery’s artworks database. It’s built on top of the concept started by the Send Me SFMOMA service and, like that service, I can ask it to send me things from the gallery’s collection. Here I’ve asked the bot to send me a sunrise to cheer myself up, and I’ve received back virtual sunrises as represented in their art collection.
However, you need to consider how quickly you can update content when circumstances change rapidly. The opening hours and public programmes are an example of things affected by the coronavirus outbreak. Auckland Art Gallery has coped with this by providing website links in their chatbot responses, and then when the website is updated the chatbot still refers to the latest information.
That Chatbot is one of many built on top of Facebook Messenger. Messenger has a built-in understanding of natural language and it has a programming language for building up questions and answers and integrating different data sources. It’s a kind of “Choose Your Own Adventure” writing tool.
One use of Facebook Messenger that I really like is Sarah Powell’s Copyright bot. It helps museums navigate the complex workflow for determining the copyright of a work in New Zealand by presenting this as a simple question and answer chat.
Bots can provide a one-way method of sharing content. This bot uses Twitter’s programming interface to share four random works from the Rijksmuseum every day. Bots like these can help surface content that might not normally get the attention of the wider public. Here for example we see the lovely illustration of squid from the 1700s.
This field is being pushed forward in several ways. Some bots are able to use unsupervised learning to improve their conversation skills. Some can be voice driven, like Siri and Alexa. Some are context aware, so for example showing you more useful responses based on your current location. Others use more specific data sources, like this example called Woebot. It’s a therapy bot with lots of built in resources for helping people deal with issues such as stress.
In 2018 the Smithsonian introduced Pepper the Robot. Pepper is still working at the museum and combines several elements – a moving mechanical robot with built-in scripted conversations and a few dance moves, along with a tablet screen for displaying image and video responses. It does provide a novel way of engaging with visitors, but we’re also entering uncanny valley as the bots get better at mimicking us.
There are ethical considerations with the use of bots. Will users be comfortable sharing their location information? Are you ok with relying on digital giants like Facebook and Amazon? What happens to the data that the bot gathers?
Chatbots also only work if they can respond to a reasonable proportion of typical questions, so test it well before you make go live. I asked Siri “Open the airlock HAL” and found, perhaps unsurprisingly, that she was unwilling to talk about the subject.
Plan out the decision tree and think about what happens if the bot gets to a dead-end. You don’t want a situation like the Personal Reminder chatbot and the Endless Screaming chatbot who were stuck together in a loop.
If you’re building a chatbot and don’t want to do coding, then you should use an out-of-the-box platform. There are several well-developed systems where you can quick build up possible conversation flows without writing any code. These platforms include Chatfuel and Botsify.
Here’s the Guggeheim-Art Tumblr account, a bot built independently of the Guggenheim. It posts daily links for works from the Guggenheim’s online collection.
Trying out different bots has reinforced one thing for me. The images you create, and the online collections you share, provide a foundation for many ways of engaging with your audience that aren’t tied to your physical venues. The more accessible your content is and more the open the licences are for re-using the content, the more likely it is that the content will be used in new ways.
I’ve added a slide with the bots and tools I’ve mentioned. As a bonus there’s a link to Seb Chan’s conference themed “Choose Your Own Adventure”. I’ll be uploading the slides to Slideshare.
That’s a lightning overview of bots. Hasta la vista!