AI can write the post.
It can't be the person.
Generative AI has changed what it means to "participate" online. When a student can paste an assignment prompt into ChatGPT and get a polished discussion post in seconds, the entire premise of text-based online engagement is compromised. Not just the grading, but the learning itself.
The detection arms race is unwinnable
The response from most of edTech has been an arms race: AI generates content, then other AI tries to detect it. Students find bypass tools. Detection improves. Bypass tools improve. The cycle continues, and the thing that's lost in the middle is the actual learning that participation was supposed to produce.
We think the detection approach is structurally flawed because it treats the symptom rather than the cause. The cause is that most online interaction formats are built on outputs that AI can easily replicate: polls, quizzes, text discussion boards, written reflections. If the medium allows it, students will use AI. Not because they're lazy, but because the format makes it the path of least resistance.
The problem isn't that students are cheating. The problem is that we're asking them to perform tasks that a machine can perform better.
The solution isn't better policing. It's better formats: interactions where the human skills are the interaction, and where authentic participation is easier than trying to fake it.
Format, not detection
When students respond in VoiceThread, they use their voice and video. They annotate content in real time, pointing at a specific element while explaining their thinking, responding to a peer's argument while referencing shared media, demonstrating a technique while narrating their process.
The interaction is multimodal and simultaneous: speaking while pointing, explaining while demonstrating, reacting while listening. This simultaneity is what makes the format resistant to AI shortcuts.
Current AI cannot produce a real-time voice response that annotates specific visual content while maintaining the social and intellectual coherence of an authentic conversation. The cognitive processes involved — integrating visual attention, verbal expression, social awareness, and domain knowledge in real time — are precisely the skills the interaction is designed to develop. They can't be offloaded without eliminating the learning.
In VoiceThread, authentic participation is the path of least resistance. It's easier to just do it than to try to fake it.
No AI detection tools are required. Not because we've found a clever workaround, but because the format itself makes authentic human presence the default.
What this means for your institution
Faculty
Assign meaningful discussions, presentations, and peer interactions knowing that students are developing real communication skills: speaking, listening, critical thinking, empathic response. Not practicing their ability to prompt an AI.
Administrators & IT
No additional AI detection services needed on top of VoiceThread. The format addresses integrity at the interaction level, not the policing level. This simplifies your technology stack and your academic integrity conversations.
Students
Research consistently shows students prefer VoiceThread's format to text-based discussion boards. They report that hearing their peers improves their learning. Over 3,337 research articles cite VoiceThread's impact on teaching and learning.
How VoiceThread uses AI
We believe in being direct about this.
Our position
AI is a powerful tool, and it's here to stay. The question for education isn't whether students will use AI — they will — the question is whether the learning environments we build can still develop the human skills that matter.
Text-based online interaction is increasingly unable to do that. Not because of any failure of the tools, but because the format itself is now vulnerable. The skills that text-based discussion was supposed to develop (articulation, critical thinking, engagement with peers) can now be performed by a machine.
VoiceThread's position is simple: build interactions where the human skills are the interaction. Where speaking, listening, and connecting with others aren't just the medium — they're the outcome. In a format like that, AI doesn't need to be detected or banned. It's just irrelevant.
Read the Research · Contact us