UiA professor Ole-Christoffer Granmo’s Tsetlin machine is part of the new exhibition at the Norwegian Museum of Science and Technology.
This is a permanent exhibition, and it was opened by HRH Crown Prince Haakon in Oslo on 7 April.
Among the 350 objects in the exhibition, the telegraph machine from 1908 is the oldest and Ole-Christoffer Granmo’s Tsetlin machine is the youngest.
“This is great fun. It is an exhibition that not only looks back but shows the importance of artificial intelligence in the future”, says professor at the University of Agder (UiA), Ole-Christoffer Granmo.
Granmo heads UiA’s Centre for Artificial Intelligence Research (CAIR) and is the man who developed the Tsetlin machine, which was first launched in 2018.
When he launched the invention, Granmo said that it would be a pioneering breakthrough for artificial intelligence if new research confirmed his findings. Today researchers from all over the world have confirmed that the Tsetlin machine is faster, simpler and more precise than the artificial intelligence that uses so-called deep learning.
The Tsetlin machine is not a machine. It is code used by artificial intelligence. In the exhibition, it is represented by a newly created computer chip, and is introduced as follows by the Norwegian Museum of Science and Technology:
The Tsetlin machine is not a machine, but a logical code to train an artificial intelligence in a computer, a machine learning algorithm in other words. What is revolutionary about the Tsetlin machine is that it is based on logic as opposed to number-based machine learning. This makes it faster, more cost effective and more accessible. The chip is currently used for various tasks in areas such as health care, finance and research.
“Tsetlin is based on logic and represents something completely new in artificial intelligence. The Tsetlin machine dismisses the artificial intelligence with deep learning that is used by Google, Facebook, Amazon, Apple and Microsoft”, Granmo says.
Artificial intelligence from the five technology giants counts the number of instances of various phenomena. But it cannot explain why, for example, women and people of colour tend to score low on rankings.
“The reason is, in short, that this type of artificial intelligence only shows statistics and superficial correlations. It cannot show context and provide understanding”, Granmo says.
This is how he describes the major companies’ use of artificial intelligence with deep learning:
“It is a machine learning method that is trained to give the right answer by fine-tuning millions of adjustable parameters. The number of parameters and the connections between them become so complicated that the calculations do not make sense to humans. The answer that is calculated can thus not be explained. Since there is a lack of transparency, the method is called black box.”
Granmo calls artificial intelligence with deep learning parrot technology. He is not alone in that. It has become a common nickname among researchers. Even scientists at Google have joined the criticism of the parrot technology used by the major companies and have since been fired.
“Parrots can imitate human speech but not understand”, Granmo says.
He dislikes that the big technology companies are developing an extremely expensive and resource-intensive technology. It is also exclusively owned by the major players. Granmo’s vision is for the Tsetlin machine to be a counterweight.
“The Tsetlin machine must be a democratic artificial intelligence. It’s free and anyone can use it. It shows context and can be scientifically verified. The tech giants’ artificial intelligence cannot because it is too complex for human comprehension”, he says.
Curator Henrik Treimo at the Norwegian Museum of Science and Technology curates the exhibition which is called I/O, which stands for input/output (Norwegian only).
“Our vision for the exhibition is for people to learn about the telecommunications and computer technology that has helped shape our society in the past and present, and which will help shape us in the years to come”, Treimo says.
He thinks the exhibition will encourage reflection and discourse about important topics related to the history of telecommunications and data. The exhibition is driven by artificial intelligence. Visitors get personal tours from artificial intelligence after answering some questions.
“Based on the individual answers, a tour is given presenting objects that the artificial intelligence thinks the person might be interested in. When you change your preferences, the path that the artificial intelligence sets for you also changes”, Treimo says.