Is it pos­sible to gen­er­ate an epis­ode in the style of an estab­lished com­edy pod­cast using vari­ous AI tools and pro­duce a res­ult that sounds as authen­t­ic as pos­sible? Driv­en by this ques­tion, the pod­cast pro­ject was born: a col­lab­or­a­tion between the enter­tain­ment pod­cast DAS PODCAST UFO and a team from the RHET AI Cen­ter at the Uni­ver­sity of Tübingen.

The project

The emer­gence of gen­er­at­ive AI tools presents new chal­lenges to our soci­ety: when we engage with media con­tent and news, the increas­ing pres­ence of AI-gen­er­ated con­tent con­stantly con­fronts us with ques­tions about the authen­ti­city of the images, videos, audio clips, and voices we encounter online, as well as the inten­ded pur­pose of this AI-gen­er­ated con­tent. Where does AI-gen­er­ated con­tent sim­u­late altern­at­ive real­it­ies that we do not imme­di­ately per­ceive as such? What con­tent can we still trust and how do we nav­ig­ate this changed media land­scape safely and with aware­ness? Our pro­ject is rooted in the ten­sion between these ques­tions and chal­lenges. As a research cen­ter for rhet­or­ic­al sci­ence com­mu­nic­a­tion on arti­fi­cial intel­li­gence, we see it as our respons­ib­il­ity to find answers to these ques­tions in dia­logue with soci­ety and to provide guid­ance in the sur­round­ing discourse.

In 2024, we were con­tac­ted by the hosts of DAS PODCAST UFO about a col­lab­or­a­tion: togeth­er, they wanted to cre­ate an epis­ode gen­er­ated entirely by arti­fi­cial intel­li­gence in their own pod­cast­ing style. From a sci­ence com­mu­nic­a­tion per­spect­ive, this was a very excit­ing pro­ject for us: The pro­ject aims to explore the lim­its and pos­sib­il­it­ies of cre­at­ive AI gen­er­a­tion and co-cre­at­ive inter­ac­tion between humans and AI tools, and to fur­ther stim­u­late the dis­course sur­round­ing the use of gen­er­at­ive AI: Where does the use of gen­er­at­ive AI tools make sense, and where does it not? How can con­sumers recog­nize AI-gen­er­ated con­tent? What does respons­ible use of gen­er­at­ive AI look like? How are view­ing, listen­ing, and read­ing habits chan­ging due to these tech­no­lo­gies? These are all ques­tions that we as a soci­ety must ask ourselves, dis­cuss pub­licly, and research scientifically.

At the RHET AI Cen­ter, we see this pro­ject as a unique oppor­tun­ity to research the field of co-cre­ativ­ity and the aes­thet­ics of impact in AI-gen­er­ated media con­tent across all stages of pro­duc­tion and dis­tri­bu­tion, and to test the viab­il­ity of rhet­or­ic­al strategies when deal­ing with gen­er­at­ive AI. At the same time, this pro­ject aims to raise pub­lic aware­ness of the social dimen­sions of AI-gen­er­ated con­tent and to foster open dis­cus­sion on these issues. In par­tic­u­lar, we are deeply com­mit­ted to the top­ic of com­pet­ent and respons­ible use of AI and AI tools, the so-called AI lit­er­acy. The pod­cast pro­ject offers an oppor­tun­ity to explore this top­ic in a non-aca­dem­ic, enter­tain­ment-ori­ented con­text togeth­er with the hosts and the PUFO com­munity. On Monday, March 16, 2026, listen­ers of DAS PODCAST UFO had the oppor­tun­ity to ask the research­ers ques­tions about the pro­ject and the research dur­ing an "Ask Me Any­thing" ses­sion hos­ted on the podcast’s own Red­dit page.

We are happy to have the enter­tain­ment pod­cast DAS PODCAST UFO and its two hosts, Stefan Titze and Florentin Will, as pro­ject part­ners by our side, with whom we can con­duct this social exper­i­ment and thus per­form import­ant pion­eer­ing work in the Ger­man-lan­guage AI discourse.

Goal and structure of the research projekt

The goal of the research pro­ject was to cre­ate a pod­cast epis­ode of PODCAST UFO that soun­ded as authen­t­ic as pos­sible using vari­ous AI tools. Anoth­er pod­cast epis­ode, in which the pro­ject is ana­lyzed, con­tex­tu­al­ized, and explained, was part of the design from the very beginning.

The entire pro­cess of the pro­ject is of interest to our work at the RHET AI cen­ter. Which cri­ter­ia does an AI-gen­er­ated epis­ode need to meet to sound authen­t­ic? How can we train gen­er­at­ive AI and design a stra­tegic prompt­ing pro­cess so that the out­put is con­vin­cing and authen­t­ic? How much human influ­ence does this pro­cess require, and at which points is this influ­ence par­tic­u­larly neces­sary? How does an epis­ode like this affect listen­ers? Where does the illu­sion break? How do listen­ers react to the epis­ode after being informed of its AI-gen­er­ated nature, and what dis­course does the pro­ject spark? These are some of the ques­tions that arose dur­ing the course of the pro­ject and which we intend to address in our research.

The work on the pro­ject began in Janu­ary 2025 and pro­ceeded in small steps until the release of the AI-gen­er­ated epis­ode on March 11, 2026, the sci­entif­ic obser­va­tion of the fol­low­ing dis­course, and the release of the explan­a­tion epis­ode on March 12, 2026.

AI tools used

Dur­ing the pro­duc­tion of the pod­cast epis­ode, our goal was to gen­er­ate as many ideas, as much text, and as much audio mater­i­al as pos­sible with AI tools. To achieve this, we divided the work in four parts and selec­ted a suit­able AI tool for each of these steps. We began by tran­scrib­ing the podcast's audio con­tent to cre­ate a text data­base, which we then used to train anoth­er AI tool to gen­er­ate the episode's text. Finally, we used a third AI tool to gen­er­ate the audio con­tent for the epis­ode. In addi­tion, a fourth AI tool was used to cre­ate a listener's style intro typ­ic­al of the podcast.

The creation process

Work on the pro­ject and the cre­ation of the final product "Exper­i­ment" took place in close col­lab­or­a­tion with our pro­ject part­ners from DAS PODCAST UFO and unfol­ded in three phases. Dur­ing each of these phases, we con­tinu­ously eval­u­ated our res­ults with Stefan Titze and Florentin Will and adjus­ted them based on their input.

The podcast episodes (in german)

• AI-gen­er­ated epis­ode: Exper­i­ment
• Res­ol­u­tion/Be­hind-the-scences epis­ode: UFO506 KI-Exper­i­ment Auflösung

Information about the podcast

Out­side of DAS PODCAST UFO Florentin Will (left) and Stefan Titze (right) also reg­u­larly per­form togeth­er on stage as improv com­edy artists. (CC: Joseph Strauch) 

Academic publications

Köhler, Anna; Volz, Car­olin; Gott­schling, Markus (2025): "Wie echt klingt KI?" In: Bieck, Julia; Staves­and, Meena: 10 Jahre HOOU. 10 Stim­men zum Pod­cast­ing. S. 59–69. Avail­able online (in ger­man): Digital-HOOU-Podcasts-01-Druckbogen.pdf.

Köhler, Anna; Volz, Car­olin (2026): "(Re-)Creating a com­edy pod­cast using Gen­er­at­ive AI." Presen­ted at the Sym­posi­um Team­ing Up with Gen­er­at­ive AI: From Tool Use to Part­ner­ship. Deutsches Hygiene-Museum Dresden, Ger­many. March 25 — 26, 2026.

Volz, Car­olin (forth­com­ing): Sim­u­la­tion and Authen­ti­city — Rhet­or­ic­al Strategies with­in the AI gen­er­ated Pod­cast (work­ing title). Mas­ter­thes­is in the dre­gree pro­gram Allge­meine Rhet­or­ik (gen­er­al rhetoric).

Projectlead and contact:

Vladimir Jakimenko (02/25 — 05/25)

Audio Pro­duc­tion; Audio Edit­ing; Research

Alina Habermann (06/25 — 08/25)

Audio Edit­ing

Impressions from the projekt

A final Thank You

Huge thanks go to Stefan Titze and Florentin Will from DAS PODCAST UFO for giv­ing us the oppor­tun­ity to col­lab­or­ate on this pro­ject and for provid­ing us not only with their data and voices but also with the plat­form to con­trib­ute to this import­ant social discourse.

We would also like to thank the KI-Maker­space in Tübin­gen for provid­ing us with space and equip­ment to record the res­ol­u­tion episode.