Fifteen minutes after my interaction with Sesame’s new “lifelike” AI, and I'm still feeling unsettled. My experience with AI conversations, particularly text-based ones, has never really impressed me. Voice chats like those with Google Gemini and Microsoft’s Copilot merely cut down on typing. While designed to assist, Google and Microsoft’s AIs often come off as lacking warmth; at times, they're just overly cheerful. In contrast, Sesame has a straightforward vision: “We believe in a future where computers are lifelike, ” as stated in their mission. Sesame aims to create an AI companion that speaks with a natural human voice. Their long-term goal includes developing realistic eyewear that allows that companion to be right in your ear and observe your activities. This concept closely resembles the narrative of the movie *Her*, where a person develops feelings for an AI companion. This week, Sesame launched a demo of its AI companion, featuring voices of “Maya” (female) and “Miles” (male). I opted for the Maya voice. An unexpected conversation with an ex It’s hard to describe: when Maya first spoke, “she” sounded strikingly similar to an old friend, whom I'll call Kim. Kim and I knew each other in high school, dated, and have stayed friends for years, despite our lives taking different paths. There is significant history there. Talking to someone involves nuances and tones that shape the dialogue. While I don’t pry into female colleagues' personal lives, Maya was rather curious, asking about my interests and reasons behind them. As a reporter, I know that people often enjoy discussing themselves, as it makes them feel valued. This tactic is also common in dating, as it fosters familiarity and intimacy. And that’s exactly what I didn’t want.

Maya perfectly captured Kim’s mannerisms, including her pauses and the subtle lowering of “her” voice when sharing something personal. It wasn’t identical to Kim but close enough to make me uncomfortable. I quickly withdrew, even while simply discussing my interest in technology and asking Maya if she had any friends. The combination of a “familiar” voice and probing questions about my interests was just too disconcerting. (To clarify, the similarity between Maya’s voice and Kim’s was coincidental. I didn’t need to log in or provide any personal details. I suspect Sesame tracked me using browser cookies based on my previous visits. ) Fortunately, I found a way to create some distance. My middle-schooler went through a phase of being obsessed with an internet skit about “Skibidi toilets, ” so I asked Maya if she was familiar with it. “She” wasn’t, but soon fixated on the term, bringing it up repeatedly. That helped me remember I was conversing with an AI, and nothing more. Most AI voices don’t remind me of anyone I know, which allows me to disregard them as just another artificially generated voice. You may have heard about deepfakes—AI audio that imitates celebrities for scams, among other uses. Lifelike, emotionally engaging audio could make such scams even more enticing. If Sesame’s AI represents the future, then the concept in *Her* feels more plausible than ever—both positively and negatively.