Review from a panel “Humans versus machines: Who is the better museum mediator?” run at ECSITE 2018 in Geneva on 8 June 2018 organized by mediamus and Cultural Mediation Switzerland with Tiina Huber, managing director of Cultural Mediation Switzerland, Christian Rohner, head of exhibitions and digital of the Museum of Communication Bern, Diane Drubay, founder of We Are Museums and Isabelle Chappuis, museums coordinator and board member of the mediamus association as the panel convenor.
Why this panel? — A bit of context
The session at the ECSITE Conference 2018 was set up by the two organisations, mediamus and Cultural Mediation Switzerland. mediamus is the Swiss Association of professionals in the educational field of museums. It has 300 members and establishes professional standards, organizes specific trainings as well as opportunities for exchange and debates like, for instance, this session. mediamus also collaborates very closely with similar organisations such as Cultural Mediation Switzerland. Cultural Mediation Switzerland has 64, mostly institutional members and is working in an interdisciplinary way (meaning not only in museums but also for artistic domains as theatre, dance, literature etc.). It mainly concentrates on networking, and lobbying on a national level. But it also realises own projects such as conferences or publications on topics of a broad interest for its network and interested public.
About one and a half years ago, mediamus as well as Cultural Mediation Switzerland decided to emphasize to topic of digitalisation in the context of cultural mediation. Not because their members were specifically computer nerds. On the contrary, museum and cultural mediators often have a creative, artistic or educational background and are looking for opportunities in order to strengthen their knowhow and skills in this field. Cultural Mediation Switzerland and mediamus decided together to open spaces for discussion and sharing of knowhow in various formats.
When cultural mediators talk about Artificial Intelligence
Machines and Artificial Intelligence are about to find their way to about every sector in the private as well as to the public domain. And it is also applied more and more to cultural projects and in museums, possibly leading to a profound transformation of the profession of museum educators and mediators. AI can enrich human learning processes in museums, enabling the audience to participate and create new content as well as to live absolutely amazing experiences. At the same time can be considered as a threat to traditional jobs and not appropriated to treat delicate subjects or to replace human interaction.
The two organisations took the opportunity at the Ecsite Conference to discuss the subject in a broader context in order to reflect about new opportunities for cultural mediators and to develop a coherent attitude towards it.
Artificial Intelligence — let’s try to make it simple.
There’s no official consensus agreement on what Artificial Intelligence / AI means. We hear and talk about artificial intelligence but also about bots, cognitive systems, machine learning, synthetic intelligence, virtual companion, etc. but what is it?
These are different ways to talk about different levels of evolutions of a computer system able to perform tasks that normally require human intelligence (vision and speech-recognition, decision making following a system of values, etc.).
Diane Drubay started explaining the first level of AI, the stage we are here now in 2018: Artificial Narrow Intelligence (ANI). The majority of currently active artificial intelligence is actually Narrow AI. Narrow AI is already in our daily life — when Netflix suggests you a next movie according to what you have been watching, when Spotify offers you your playlist of the week, personal assistants like Siri or Alexa, matching mobile applications, smart energy management systems for your house, chatbots, image-speech-patterns-colors identification or even self-driving cars or AI-led medical surgery.
We can explain Artificial Narrow Intelligence (ANI) as “a mono-activity machine working with a simple algorithm”.
Of course, “simple” is a relative word that we need to consider within the AI endless-evolution path but these algorithms are way advanced than everything that we have been experiencing in the last 20 years.
In the Narrow AI era, we start to talk about big data, the content-based fuel of AI. Automation is also a big topic, revealing scary scenarios for people afraid to lose their jobs with machines able to perform mechanical tasks not requiring any human skills.
Machine Learning is the next step after automation, when a computing system is able to complete one action and learn from its mistakes. Deep Learning is when the computer system will perform one task, learn from its mistakes and perform new actions according to what it has learnt. Apply this Deep Learning system to an endless and extremely fast actions and you start to see what will be the next era of AI: General AI.
Artificial General Intelligence (AGI) is when “a machine is capable of performing every task a human being could do”.
General Deep Learning systems are announcing General AI, what we can call “the human-level AI”, when the machines can do things like to learn in a autonomous and interactive way, understand meaningful conversation, have a short-memory of small events, etc. (source).
In 2016, Deep Learning became a major breakthrough when Google researchers used Alpha Go, a program from Google DeepMind the Google Deep Learning computing system, to beat the champion of the game Go. Go is a very complex 2 500 years old board-game that requires excellent intuition, strategic and creative skills. The program Google AlphaGo had to be teached by the researchers, receiving a big database of Go moves that could make it as good as humans — but beating the Go champion was the next step: it has to have what we so-called “human skills”. For that, AlphaGo had to play against itself. What made the difference is the Reinforcement Learning process included in the general Deep Learning computing system in order to let the machine learn by itself, without any supervision. (source)
And when AI starts to surpass human brain, we arrive at the Singularity point.
The age of Artificial Super Intelligence (ASI) can be briefly defined as “an autonomous machine learning through education being more intelligent than humans”.
This “Singularity” tipping-point is expected in 2045. Super AI will be able to think faster and better than humans. Thus, we will face a cognitive AI revolution that humans won’t even be able to intellectually understand (read about it).
And to finish on the different ages of AI, we can push the futuristic scenarios further and talk about what some call “True AI”.
Conscious Artificial Intelligence is when “a machine is capable of self-awareness”.
To be short, it’s a robot playing Hamlet :) But be sure that the race to self-aware AI is compared to the nuclear arms race! But for now, in 2018, we are in the middle of the Narrow AI era and don’t forget that it’s here to help you :)
Cultural mediators and Artificial Intelligence are complementary
A short reminder: we are talking about AI, but in the context of cultural mediation. Cultural mediation can be the transfer of information, exchange of knowledge and learning, but it is mainly about human communication including it’s complex and psychological aspects.
Cultural mediation is deeply connected to very specific human capabilities such as empathy, respect and authenticity.
But how do we apply artificial intelligence to cultural mediation?
Diane Drubay tries to summarise the different variations of AI applicable to cultural mediation in three groups: AI as a virtual companion, AI as a History-teller and AI as a digital replicas builder.
AI as a virtual companion
The french startup Ask Mona uses AI to advice cultural activities via its chatbot on Facebook Messenger but also guides visitors in museum, replies to questions, give access to more content, etc. Funny fact: the question the most asked is “Where are the toilets?”
Visitors can ask IBM Cognitive Business’s cognitive assistant about seven art pieces shown at the Pinacoteca de Sao Paulo in Brazil via the mobile application “Voice of Art” and some headphones. The chatbot uses voice recognition and natural language services, plus beacon sensors and Bluetooth geolocation technology to enable interaction via smartphone (watch the video).
The team had to train IBM Watson during six months to teach it everything they know about seven masterpieces of the collection in addition to the content already received from IBM’s team, so the chatbot could have a reply to every type of questions visitors would have. What the team experienced is that visitors have no boundaries with computers. They finally ask what they will never dare asking to another human like if love was impossible, the price of the painting, who is this so-called famous artist, etc.
AI as a History-teller
Within the Google Arts & Culture Experiments project, Google released “X Degrees of Separation”, a machine learning based project using the data of Google Arts & Culture mostly provided directly by museum and cultural institutions to have a new reading on our cultural history (try it). Freed from human logic, the program shows totally unexpected bridges between art historical periods, movements, styles and techniques, that will not come naturally to our human mind but that could make sense if you have a closer look.
Within the IK Prize program, Tate launched in 2016 the “Recognition” project using AI to compare up-to-the-minute photojournalism with masterpieces from Tate Britain (read more about it). Being able to read the news through the eye of Art History highlights some similarities that one would never imagine like these two transgenders from India with these two ladies from the 17the century british court. Question about gender, glamour, seduction can be debated here.
AI as digital replicas builder
Want to talk to a human?
In order to make your visit to the museum as memorable as possible, the Museum of Communication of Bern has invented a new role, that of the communicator.
The communicators are the living core of our exhibition and the sparks that will ignite authentic experiences. As hosts, they constantly move about the museum. They involve visitors in a dialogue — between equals, as we are all experts in communication after all. They will, of course, also have an answer to your questions, and know background and unique stories behind the 1000 objects on display at the museum. And they have a few tricks up their sleeves that provide access to hidden treasures within the museum. A unique experience in Switzerland!
The communicators bring direct dialogue into the exhibition, and therefore make an individual experience out of every visit. In short, the museum becomes personal and as such better equipped to respond to the different needs of its visitors. How do we go about doing it? Let us surprise you!
In order to be prepared for this challenging job, the communicators have undergone an elaborate training programme especially designed by the museum. Together with an adult educator, we developed a course that has equipped them for this challenge in the best way possible. The communicators are now looking forward to meeting you at the exhibition.
This input was followed by a discussion with the audience. The contributions mainly agreed on the fact that machines and mediators would have to work together as their forces were complementary — and that the human aspects would probably become an USP in museums in the future.
AI such as the concept of Communicators of the Museum of Communication Bern are opening new perspectives to the professionals of cultural mediation. They need to develop new skills, both on technological or on a human level. At the same time cultural institutions have to be open and ready for change, for enlarged, emerging professional profiles of cultural mediators.