I’ll never forget the demo. We were pitching MovieLaLa to a room full of investors, and our killer feature was a voice-powered movie recommendation engine. In my head, it was going to be like talking to a super-smart film buff. In reality, it was like talking to a bored robot who’d just read the back of a DVD case.
“I… am… searching… for… movies… like… ‘The… Godfather’,” it droned. The room was silent. I could feel the investors’ enthusiasm draining away with every stilted word. We didn’t get the funding that day. And while MovieLaLa was eventually acquired by Gfycat, that experience taught me a valuable lesson: when it comes to voice, “good enough” is never good enough.
That was years ago, and the world of AI voice has come a long way since then. We’ve gone from clunky, robotic voices to ones that are almost indistinguishable from humans. But here’s the thing: they’re still not quite there. We’re stuck in the uncanny valley of voice, and it’s a frustrating place to be.
As a serial entrepreneur and angel investor, I’ve had a front-row seat to the AI revolution. I’ve seen companies like Anthropic, OpenAI, and Hugging Face push the boundaries of what’s possible. I’ve also seen countless startups crash and burn because they underestimated the complexity of building a truly human-like AI voice. It’s a problem that’s fascinated me for years, and I believe we’re on the cusp of a major breakthrough.
The Soulless Pursuit of Perfection
One of the biggest mistakes I see in the AI voice space is the obsession with perfection. Everyone is chasing a voice that is perfectly clear, perfectly paced, and perfectly free of any “ums” or “ahs.” But here’s the secret: perfection is boring. It’s the imperfections that make us human. It’s the slight hesitation in our voice when we’re not sure about something, the way our pitch rises when we’re excited, the way we sometimes stumble over our words. These aren’t bugs; they’re features.
I remember when we were building RemoteTeam, which was later acquired by Gusto. We were creating a tool to help distributed teams feel more connected, and we experimented with using AI-powered avatars for virtual meetings. We thought that if we could create perfectly realistic avatars with perfectly clear voices, it would be the next best thing to being in the same room. We were wrong. The avatars were so perfect that they were creepy. People felt like they were talking to robots, not their colleagues. We ended up scrapping the idea and focusing on more practical tools for remote collaboration.
More Than Just Words
The Turing Test is obsolete. We need to move beyond just creating AI voices that can fool us into thinking they’re human. We need to create AI voices that can connect with us on an emotional level. This is the holy grail of AI voice, and it’s a problem that’s far more complex than just stringing together a bunch of phonemes.
Think about all the subtle cues we use to convey emotion in our voice: the pitch, the tone, the rhythm, the volume. These are all things that are incredibly difficult for an AI to replicate. But we’re starting to see some progress. Companies are using generative adversarial networks (GANs) to create AI voices that are much more expressive and natural-sounding. They’re also experimenting with using sentiment analysis to train AI voices to respond to the user’s emotional state.
The Future is a Conversation
I believe the future of AI voice is not a monologue, but a dialogue. It’s not about creating a perfect, all-knowing oracle that can answer all our questions. It’s about creating a conversational partner that can learn from us, adapt to our needs, and even challenge us. It’s about creating an AI that we can have a real conversation with, not just a series of commands and responses.
I’m putting my money where my mouth is. I’ve invested in over 200 companies, and a growing number of them are working on some aspect of AI voice. I’m particularly excited about companies that are focused on creating AI voices for specific applications, like education, healthcare, and entertainment. I believe that by focusing on a narrow domain, these companies can create AI voices that are not only more realistic, but also more useful.
If you’re an entrepreneur working on something in the AI voice space, I want to hear from you. I’m not looking for another Siri or Alexa clone. I’m looking for something new, something different, something that will finally get us out of the uncanny valley and into a world where we can have truly meaningful conversations with our AI companions. The future of voice is a conversation, and I’m ready to start talking.
Frequently Asked Questions
How has this view evolved over time?
My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.