In a matter of seconds, AI models like ChatGPT and others provide us with answers to every conceivable question. But can we trust these answers? Five international student teams have developed solutions for such and other challenges surrounding the use of artificial intelligence. In the hybrid master’s seminar “Ethical Innovation and AI” at the Institute for Ethics in Technology at the Hamburg University of Technology (TUHH), they created their own software applications that combine ethics and artificial intelligence: from detecting AI-generated content to environmentally friendly prompts and combating gender biases in the job application process. The teams recently presented their prototypes at the International Conference on Ethics Education in Amsterdam.
From Ethical Theory to Ethical Practice
The master’s seminar was designed and conducted by Dr. Jonas Bozenhard from the Institute for Ethics in Technology. “The seminar follows a prototyping-based approach,” explains Bozenhard. “In teams, the students developed prototypes that address ethical challenges arising from modern generative AI.” Instead of merely discussing AI risks, the student teams built their own functioning software systems. These follow the principle of “Ethics by Design,” meaning ethical standards such as transparency and user control are considered from the very first line of code. “As AI systems become increasingly powerful, they raise profound philosophical questions,” says Bozenhard. “A central goal was to foster an understanding that technical design decisions have far-reaching ethical implications – we want to create a bridge between ethics and engineering.”
The seminar was also offered online as an official ECIU Challenge (European Consortium of Innovative Universities), making it internationally accessible – students from five other European universities participated alongside TUHH. Finally, the seminar participants traveled to Amsterdam for the International Conference on Ethics Education. The trip and participation were made possible by an ECIU grant secured by Bozenhard. In Amsterdam, the students had the opportunity to present their work on an international academic stage. An overview of the five prototypes:
“AI Content Checker”: Automatically Detecting AI-Generated Content
A central problem of modern generative AI is the declining trust in digital content. The “AI Content Checker” prototype addresses this issue. The application is simple to use: users upload a text, the system analyzes linguistic patterns and then outputs the text sentence by sentence with color-coded markings: green marks indicate human content, red marks indicate AI-generated content. For each sentence, a percentage indicates the likelihood that it is AI or human content. Based on analysis of text complexity and sentence rhythm, the system also transparently explains why a passage was classified as AI-generated.
“Ethical Analyser”: A Lie Detector for AI Texts
Taking a step further, the “Ethical Analyser” prototype evaluates the credibility of AI answers via a browser extension. This is especially relevant because large language models often answer confidently but frequently fabricate facts and “hallucinate” information. The analyser checks AI responses for source reliability, semantic similarity, and emotionally manipulative language. Whenever ChatGPT, Gemini, or Claude generates a response, a pop-up appears automatically, providing a credibility score explained from 100 (very credible) to 0 (completely not credible). Although the tool does not yet offer complete protection against misinformation, it can be an important aid in assessing AI answers.
“EPOCH”: Prompting AI More Efficiently
The students in Amsterdam also addressed the invisible ecological costs of AI technology. The “EPOCH – GenAI Efficiency Engine” prototype, also a browser extension, tackles the massive energy demands of large language models. The longer and more complex the queries posed to AI, the greater the consumption of energy and water. To reduce this, EPOCH activates automatically before a prompt is sent and suggests a shorter alternative prompt. The prototype also indicates the expected CO₂ savings, encouraging shorter and thus more environmentally friendly queries – following the motto “Nudge, Don’t Force.”
“FairCV”: Better Opportunities for Female Applicants Thanks to a CV Tool
That AI can also dismantle societal barriers is demonstrated by the project “FairCV.” Studies show that women tend to undervalue their achievements in resumes and choose more defensive phrasing. FairCV breaks this cycle. Using precise prompt engineering, the tool optimizes resumes to minimize gender-specific biases in the recruitment process – without rewriting genuine qualifications.
“ETHOSDrive”: Artificial Intelligence as a Life Saver
While tools like “FairCV” reduce societal barriers, the “ETHOSDrive” project addresses extreme situations in road traffic. When an autonomous vehicle fails or crashes, the passengers experience massive stress. The students’ solution: an empathetic crisis co-pilot powered by AI. The system locally stores 27 specific trigger words such as “crash” or “bleeding.” If a passenger utters any of these triggers, a digital rescue chain starts automatically. The digital co-pilot immediately initiates medical first aid instructions according to WHO standards, alerts emergency services, and stabilizes the passengers with a calming voice output. To protect privacy in this exceptional situation, the system operates entirely without cloud dependency: all audio data is encrypted and processed locally, then automatically deleted after 48 hours.
Stepping onto the European World Stage
The congress in Amsterdam is only the beginning for the students. The concept behind the “Ethical Analyser” attracted such strong interest that the team was advised to further develop it. Experts noted that the underlying system could be excellently adapted to protecting intellectual property in the context of generative AI. The team then participated in the European IP4OS Open Innovation Challenge and emerged as winners. As a prize, the TUHH students will present their newly focused prototype in October at the IP4OS Final Policy Event of the European Parliament in Brussels.
The Minds Behind the Prototypes
“AI Content Checker”:
Tayyip Canbay, Baris Kaplan, Mihais Jersovs, Terry Lians, Divya Chidananda
“Ethical Analyser”:
Niharika Kiran, Praveen Krishnakumar Megharaj, Jesvin Varghese
“EPOCH – GenAI Efficiency Engine”:
Marcel Dietl, Meenakshi Kanchibhotla, Utkal Sanjel, Levin Plümper, Ayman Imran
“FairCV”:
Elsa Bylin, Barbara Fischer, Ehab Mansour, Muhammed Raza
“ETHOSDrive”:
Tawfiq Bashar, A N M Fariha Akhter, Elmarko Fusha, A R M Nazmul Ahsan Anupom, Shahbaz Rahman
