Fri­day, 27th of Octo­ber 2023. A chilly gar­age at dusk — the swt-Kul­tur­Werk. Nearby, the Neck­ar flows quietly and unob­trus­ively. A sol­it­ary fig­ure hur­ries across the street towards the swt-Kul­tur­Werk and dis­ap­pears inside. But sud­denly, just before 5pm, the calm pre-Hal­loween atmo­sphere is shattered: people begin to stream in from all dir­ec­tions. The swt-Kul­tur­Werk fills with a mostly young audi­ence, eager to see the IN-ML-OUT exhib­i­tion. After all, even so close to Hal­loween, we remain deeply intrigued by the top­ics of wind energy and AI.

The cent­ral ques­tion of the even­ing is:

"How can AI assist in mak­ing wind energy as effe­citve as pos­sible in times of cli­mate change ?

The entire pro­ject was made pos­sible through a col­lab­or­a­tion between the Tübin­gen Excel­lence Cluster "Machine Learn­ing: New Per­spect­ives for Sci­ence" (ger: "Maschinelles Lernen: Neue Per­spekt­iven für die Wis­senschaft"), the "State Academy of Fine Arts Stut­tgart" (ger: "Staat­liche Akademie der Bildenden Kün­ste Stut­tgart"), and the RHET AI Cen­ter. The three abbre­vi­ations in the exhib­i­tion title, "IN", "ML", and "OUT", rep­res­ent the inter­con­nec­ted exhib­its: first comes the input, in the middle the machine learn­ing, and finally the out­put, or simply: IN-ML-OUT.

The con­cep­tu­al start­ing point of the pro­ject is the research con­duc­ted by Nina Effen­ber­ger. Her dis­ser­ta­tion pro­ject, part of the research group "Machine Learn­ing in Sus­tain­able Energy Sys­tems" (ger: "Maschinelles Lernen in Nach­halti­gen Ener­giesyste­men"), led by Dr. Nicole Lud­wig at the Tübin­gen Excel­lence Cluster "Machine Learn­ing", explores how wind energy and its usage are affected by cli­mate change — and how mod­ern algorithms can help improve long-term pre­dic­tions of wind speeds.

^{©Alina Haber­mann
Nina Effen­ber­ger speak­ing}

In times of cli­mate change and the grow­ing uncer­tainty sur­round­ing the sup­ply of fossil fuels, it is becom­ing increas­ingly import­ant to make the best pos­sible use of renew­able energy sources. The chal­lenges involved, and how arti­fi­cial intel­li­gence can help to over­come them, are made tan­gible through the inter­act­ive art exhib­it "IN-ML-OUT", using wind energy as an example. Over the course of nearly six months, design stu­dents Laura Neuschel­er, Samuel Sto­ber, and Arne San­wald developed the pro­ject in close col­lab­or­a­tion with the research­ers as well as Michael Pelzer from the RHET AI Center. 

IN — Cli­mate Change: How do our actions impact the cli­mate, and why is it import­ant for energy data to be access­ible for research? 

We've all been there — you check the weath­er app in the morn­ing, and it shows a 50% chance of rain. So, you grab an umbrella, even though that 50% doesn't neces­sar­ily mean there's a 50/50 chance it will rain. Instead, it indic­ates that on half of the days with sim­il­ar weath­er con­di­tions, it rained. Nina Effen­ber­ger and Dr. Nicole Lud­wig believe that AI can lead to more accur­ate weath­er pre­dic­tions. AI can make far more sense of such per­cent­ages and the data behind them than we humans can, as "AI can assist wherever we have data that humans can no longer fully com­pre­hend", accord­ing to Effen­ber­ger and Lud­wig. They stress the import­ance of AI integ­ra­tion into weath­er fore­cast­ing to sup­port renew­able ener­gies that depend on the weath­er, such as wind power. In this field, AI works faster and more effi­ciently — a neces­sity in the fight against cli­mate change. The IN exhib­it shows the chaot­ic world of vari­ous air cur­rents, demon­strat­ing how even the smal­lest changes in cli­mate can have major effects on these airflows.

Concept and design: Laura Neuschel­er, with Arne San­wald and Samuel Stober

Sci­entif­ic con­sulta­tion: Nina Effen­ber­ger and Dr. Nicole Ludwig

ML — Machine Learn­ing: How can machine learn­ing assist in mak­ing wind energy as effect­ive as pos­sible in times of cli­mate change?

Fore­casts can become more effi­cent and accur­ate through machine learn­ing — and pre­cise cli­mate fore­casts are cru­cial for expand­ing renew­able energy. The role of AI here is to refine and enhance pre­dic­tions and to bet­ter under­stand the rela­tions­ships between glob­al cli­mate mod­els and loc­al wind energy data. The middle sec­tion, ML, trans­lates the air flows gen­er­ated in the first part of the exhib­it, "IN", into data and records them.

Concept and design: Samuel Sto­ber, with Laura Neuschel­er and Arne Sanwald

Sci­entif­ic con­sulta­tion: Nina Effen­ber­ger and Dr. Nicole Ludwig

OUT — Renew­able Energy: How have wind speeds changed over the past dec­ades, and what ini­ti­at­ives and pro­jects related to renew­able energy already exist?

Over the past dec­ades, wind speeds and pat­terns have changed sig­ni­fic­antly world­wide, and this trend is expec­ted to con­tin­ue. Storms are becom­ing stronger, and calm winds last longer than before. Accur­ate weath­er fore­casts are essen­tial to advance the energy trans­ition, as they enable the effect­ive plan­ning of new wind tur­bines. How­ever, only AI can provide these pre­cise fore­casts. AI can pro­cess and inter­pret vast amounts of weath­er data quickly and accur­ately, mak­ing pre­dic­tions for the near future. 

The final exhib­it fea­tures a world map with marked wind tur­bines. By press­ing a but­ton, vis­it­ors can jump from the past to the present and then sev­er­al years into the future to observe how wind pat­terns change and which wind tur­bines can cap­ture and con­vert these flows into energy. The inter­play of the three exhib­its demon­strates how AI can effi­ciently drive the expan­sion of new wind tur­bines and ensure that no one has to sit in dark houses, even dur­ing pro­longed calm winds.

Concept and design: Arne San­wald, with Max­imili­an Hans

Sci­entif­ic con­sulta­tion: Nina Effen­ber­ger and Dr. Nicole Ludwig

^{©Alina Haber­mann
Michael Pelzer speaking}

After the exhib­i­tion open­ing and the present­a­tion of the exhib­its, a pan­el dis­cus­sion took place, hos­ted by Prof. Dr. Olaf Kramer. The par­ti­cipants, Dr. Nicole Lud­wig, Peter Seimer and Prof. Dr. Phil­ipp Staudt, dis­cussed how AI can sup­port the energy trans­ition. Dur­ing the con­ver­sa­tion, Peter Seimer poin­ted out that while there is much talk about sol­ar roofs and wind tur­bines in the con­text of energy trans­ition, many oth­er aspects are often for­got­ten. Pro­fess­or Staudt, ref­er­en­cing the term "Smart Grid", noted that it is chal­len­ging to bring togeth­er all the applic­a­tions asso­ci­ated with it. He added that the vast num­ber of pro­cesses occur­ring in the power grid are bey­ond human com­pre­hen­sion — only AI can man­age this com­plex­ity. Nicole Lud­wig expan­ded on this idea by high­light­ing the lack of digit­al­isa­tion in the energy sys­tem, while Peter Seimer argued that digit­al­isa­tion has not yet truly arrived in Germany.

After this thought-pro­vok­ing pan­el dis­cus­sion, the buf­fet was opened with the pop of a cham­pagne bottle. The three exhib­i­tion pro­jects attrac­ted vis­it­ors with their inter­activ­ity. Guests enjoyed savour­ies, sipped cham­pagne, and explored the fea­tures of the pro­jects. A suc­cess­ful even­ing all around. …and Happy Halloween! 🎃

Start­ing from Novem­ber 16th, the exhib­it will also be on dis­play at the museum of Tübingen.