Trust can decisively influence the performance of an economic cycle. Governments and banks devote considerable resources to the field of communication to influence the level of the population’s trust. For example, "forward guidance" is one of the tools available to central banks to exercise their power in monetary policy to condition, with their announcements, the expectations of the markets on future interest rates and the cost of money. On a macroeconomic level, an increase in confidence leads to a rise in consumption and a greater propensity to invest: both fundamental components of economic growth. On the contrary, if the trust is unfounded and not supported by real development, the consequence is a disproportionate increase in demand and a subsequent unjustified rise in prices. This scenario can lead to the bursting of the so-called speculative bubbles, causing a sudden collapse of the market and consequent repercussions in social terms.

On an individual level, trust is not an innate feeling, but something that is built and evolves over time. Contact with our mothers, fathers, and the sense of security that comes from the family unit contributes to the creation of a feeling of optimism, love, protection - defined by the psychologist Erik H. Erikson as basic trust. Once this is established, it is possible for the individual to trust the outside world, believe in others, and share more complex cultural constructs such as institutions, politics, and economics.

Starting with the historical series and the analysis of economic data in real-time - representative of the current economic trend - it was possible to train a recurrent neural network (RNN) to predict possible scenarios and how trust will evolve in the future. RNN is a particular type of neural network that includes neurons linked together in a loop, allowing previous outputs to be used as inputs. They are extremely useful for processing time series in which data is analyzed sequentially. In this specific case, we used a variation of the RNN, called LSTM (Long Short Term Memory), an architecture developed to overcome problems encountered during the training of the standard RNN (Vanishing Gradient Problem).

To better develop a forecast analysis of the IFO, a monthly recorded value, we tried to relate this index to other economic parameters including the DAX, a segment of the Frankfurt Stock Exchange, which contains the 30 most key stocks. We then trained an LSTM to predict the value of the DAX, and a linear regression algorithm to bring back that value with the corresponding IFO index.

To represent the trend of trust in the past, we elaborated on a spectrum of settings ranging from the intimate and personal sphere, represented by hands and gestures, to the collective one, described through places belonging to Berlin - a symbolic city of the reunification of the two Germanys. The technique used for image processing is photogrammetry, which allows to digitally recreate representations as close as possible to reality. This gap between reality and digital re-elaboration becomes the expressive space where we endeavoured to tell how the emotional influence of trust influences our perception of the world.

During the installation, the timeline flows, reaching the present and then going beyond it. From then on, the trend of confidence is tracked on a probabilistic basis; in the same way, the visuals no longer attempt to reflect reality but begins to imagine new scenarios. A neural network trained on thousands of images of Berlin from the past creates non-existent places, of which a second algorithm tries to give a spatial, three-dimensional understanding. These moving landscapes become the place of the possible. This is a space not yet written in history, but one whose boundaries can already be imagined.

To broaden this three-dimensional understanding outside the original geographic parameters of Treu, economic data from other vibrant financial markets can be introduced into the RNN. We can effectively train the LSTM to process historical data of local markets - such as the Hang Seng, Nikkei, or Dow Jones. Utilizing the same regression algorithm to correspond with the IFO index and demonstrate trust in cultural constructs on a localized level.