Ars Electronica | Artificial intelligence at the service of people

In “Level of Confidence” from 2024, Rafael Lozano-Hemmer highlights similarities between the images of the disappeared and those who view them. Photo: Brian Davies Artificial intelligence is currently making rather negative headlines. AI can also focus on people, not just as a data resource, subscription payer and object of manipulation.

The self-learning computer programs can help improve social and cultural relationships. Language bots can transform official and legal German into easy-to-understand language. The Society for Social Business Consulting (GSUB), for example, is developing an AI-based platform that is intended to help with submitting funding applications and also relieve the authorities of the application review process. The project was presented at the “AI Impact Days,” a networking forum in the area of ​​artificial intelligence and social concerns. The Youna chatbot was also presented there. He uses sensitive language to help people cope with experiences of racism. And because it works, it will hopefully not be misused in the future as an argument for cutting funding for relevant advisory institutions.

Social AI is by no means limited to Europe. Especially in the Global South, more and more technically skilled people are using the possibilities of new technologies to solve very specific local and regional problems. You also benefit from the currently relatively low-barrier access to language models and AI agents.

In Kenya, for example, IT engineer Esther Kimani developed an AI-supported app that helps farmers identify pests in their fields in a timely manner, combat them earlier, and thus secure both food for the immediate surrounding area and income for the farmers.

Kimani herself comes from a farming background. “I am the eldest of four daughters, and as the firstborn you quickly notice when a good harvest is coming and when a bad one is coming,” she said at the presentation of her agricultural AI “Farmer Lifeline” at the “Ars Electronica” festival in Linz. Kimani was also the first young woman from her village who was able to go to the city to study. “I knew then that I was leaving my home to come back and change it,” she affirmed. On the one hand, the driving force behind the change was their knowledge of crop losses due to pests. »Every year my parents lost about a third of the harvest. And they weren't the only ones. The other small farmers had a similar situation,” she emphasized. On the other hand, while studying, she learned about the advances in detecting pests using machine learning. They finally brought the two together four years ago.

The system consists of a camera that is permanently installed in the field and continuously takes images. The energy is supplied via solar cells. The images are transferred to a specially trained AI agent and evaluated there. »The farmers then receive a warning in their language as an SMS on their smartphone. This is accompanied by suggestions on what to do about the pest or disease. “We recommend the most financially advantageous, locally available and environmentally friendly solutions, including the dosages,” explained Kimani. Kimani left it open to what extent the traditional knowledge of local farmers is included in the recommendations and to what extent the AI ​​is trained in this regard.

At the presentation in Linz, she pointed out that through the targeted use of pesticides with the help of AI, the overall use of these agents could be drastically reduced, in the case of pesticides by around 40 percent. 35 percent less fertilizer was also used. Overall, these savings since the system was introduced have also resulted in a reduction of 100,000 tonnes of carbon dioxide emissions that would otherwise have been generated during production and transport.

The system is currently used on 5,900 farms. The monthly leasing fee is three US dollars. This makes the offer affordable for small farmers too. »Other providers offer drone overflights. But it costs $100 an hour,” she said.

»Every year my parents lost about a third of the harvest. And they weren’t the only ones.”

The project also has social impacts. Farming families have more money and can invest it in education. »65 percent of the agricultural workforce in our region are women. You will be strengthened by it.”

Kimani has ambitious goals, wants to expand the project to neighboring countries, reach one million small farmers in Africa by 2030 and also ensure a noticeably better climate balance by reducing the use of fertilizers and pesticides. She also collected numerous prizes for her project, such as the Africa Prize for Engineering Innovation 2024.

The Mexican artist and trained chemist Rafael Lozano-Hemmer uses the facial recognition AI “Level of Confidence” and “Global Level of Confidence” he developed for completely different purposes. »Level of Confidence« was created in 2015.

The installation consists of a camera that records the image of people looking into the camera. In the background, facial recognition software, which is usually used by police and security forces, compares facial parameters with images of the 43 Mexican student teachers from Ayotzinapa who were first arrested by the police in September 2014 and then killed by drug cartel hitmen, although the exact course of the crime has not yet been clarified and the students' remains have not been found. The software filters out the images of the students that look most similar to the visitors to the installation. The similarity is also given in percentages.

Lozano-Hemmer is reversing the usual and widely criticized practice of facial recognition systems. Security forces usually use them to filter out suspects, which is often associated with racist attitudes. In the USA, for example, Christopher Gatlin spent 16 months in prison before it was proven that facial recognition software had incorrectly identified him as the perpetrator of an attack on security forces. Eight similar cases have now been reported in the USA, with the majority of those falsely suspected being people of color.

“Level of Confidence,” meanwhile, becomes a remembrance project for the students who disappeared and were officially declared dead four years ago. At the same time, awareness is raised of the extent of corruption and possible involvement of the Mexican authorities, the military and politicians, which became apparent in the context of the mass kidnapping and the attempted cover-up of the crime. Lozano-Hemmer is making the software for the project available as open source and encourages its reuse for other cases of enforced disappearance, such as that of indigenous women in Canada, the country where he now lives.

“Global Level of Confidence” is the artist’s own development. The data basis here is both data from Argentina's National Genetic Database and an archive of around 10,000 images of people who were kidnapped during the Argentine military dictatorship and never reappeared. It is estimated that around 30,000 people disappeared between 1976 and 1983 alone. The current camera images are compared with this data at the “Global Level of Confidence”. The 15 to 30 most similar hits are then displayed on a monitor with the opportunity to learn more information about this victim of the military dictatorship. It is a very individual remembering.

The visual comparison also has political and social explosive power. It was not uncommon for the military to take the children away from the arrested parents and give them up for adoption to couples loyal to the regime. The movement of the Grandmothers of the Plaza de Mayo assumes that 500 babies and small children have been kidnapped and given to new families. Lozano-Hemmer hopes that the installation could help adults who are now looking for their biological parents and their fate. Part of the installation also includes the possibility of comparing one's own DNA with the holdings of Argentina's national genetic database.

The data that arises from interaction with the camera is not saved, emphasizes Lozano-Hemmer. The server for the project is securely located in Iceland.

And of course there are other cases of mass disappearances and forced disappearances in which variants of the project could be helpful.

Lozano-Hemmer's double project "Level of Confidence" and "Global Level of Confidence" is characterized by the fact that it does not merely simplify processes and reduce barriers, like numerous other AI developed for the social sector. Rather, they create individual approaches to victims who usually seem far away and disappear facelessly in the mass of other victims. And access to Argentina's genetic database even gives individual victims the opportunity to reconstruct their own family connections.

As the examples show, when used in a socially conscious manner, AI can be more than just an efficiency tool and a possible means of abolishing humanity, as is currently often feared and often demonized.

Other media have corporations or billionaires behind them. We have the nd.Genossenschaft and readers like you. That makes us independent and that’s exactly how we want to remain. But it also means: without you the calculation won't work. We need 300,000 euros by the end of the year.

You can donate directly here, one-time or regularly, so that nd stays.