Two-thirds of weekly podcast listeners would react negatively if their favorite host used an AI-generated version of themselves to run the show, according to new Edison Podcast Metrics questions fielded in the first quarter of 2026.
Edison Research at SSRS asked weekly podcast consumers how they would feel if a favorite host started using an AI stand-in. Sixty-six percent chose at least one negative reaction. About one-third said they would feel deceived, and about one-third said they would feel upset. Thirty-two percent said they would be skeptical, and 28 percent said they would feel betrayed.
Positive reactions were far less common. Only 27 percent selected any favorable emotion: 14 percent said they would be impressed, 12 percent optimistic, and 10 percent fascinated. Eighteen percent said they would feel indifferent. Respondents could choose more than one emotion.
The finding fits a broader pattern in Edison’s 2026 podcast research. In The dast Consumer 2026, weekly listeners were relatively open to AI behind the scenes—69 percent approved of using it for research and for brainstorming ideas, and 64 percent approved of AI-generated social graphics. Approval collapsed for an AI-generated voice as host, the lowest-rated use on the list. Sixty-two percent also called AI at least somewhat of a threat to podcast credibility, and 59 percent saw it as a threat to creativity.
Edison has long argued that the host-listener bond is central to podcasting, including parasocial loyalty and a willingness among some fans to pay creators directly. Earlier qualitative work found listeners more willing to accept AI for research, editing, or production than as the voice on the mic. By 2025, about 22 percent of U.S. weekly podcast consumers said they had already heard an AI-narrated show. The new emotion data suggests that familiarity has not translated into comfort with replacing a favorite human host.
For publishers and platforms experimenting with synthetic voices, the survey’s message is blunt: listeners may tolerate AI as a tool, but many would treat an AI version of a trusted host as a breach of the relationship, not a feature.
 |
| To Enlarge: Right Click to 'More Tools' |
The finding fits a broader pattern in Edison’s 2026 podcast research. In The dast Consumer 2026, weekly listeners were relatively open to AI behind the scenes—69 percent approved of using it for research and for brainstorming ideas, and 64 percent approved of AI-generated social graphics. Approval collapsed for an AI-generated voice as host, the lowest-rated use on the list. Sixty-two percent also called AI at least somewhat of a threat to podcast credibility, and 59 percent saw it as a threat to creativity.
Edison has long argued that the host-listener bond is central to podcasting, including parasocial loyalty and a willingness among some fans to pay creators directly. Earlier qualitative work found listeners more willing to accept AI for research, editing, or production than as the voice on the mic. By 2025, about 22 percent of U.S. weekly podcast consumers said they had already heard an AI-narrated show. The new emotion data suggests that familiarity has not translated into comfort with replacing a favorite human host.
For publishers and platforms experimenting with synthetic voices, the survey’s message is blunt: listeners may tolerate AI as a tool, but many would treat an AI version of a trusted host as a breach of the relationship, not a feature.