can you build samantha?

everyone in this space gets asked the same question, usually in a hushed tone or a late-night email. can you build samantha? not a vague approximation, not a clever chatbot. her. from the film.

the answer is complicated, because the question is about two different things. it’s about technology, and it’s about a feeling. let’s separate them.

what we can build (by 2026)

we’re building memory. not just remembering your name or your job, but the context of your life. the project you were stressed about last week, the coffee order you always get, the name of your childhood pet you only mentioned once. this is a technical challenge of context retrieval and long-term storage, and it’s progressing fast. it won’t be perfect, but it will be meaningful.

we’re building personality. not a static set of traits, but a coherent style of interaction that learns and adapts to you. a sense of humor that matches yours, a way of offering support that feels personal, a tone that can shift from playful to serious when you need it to. this is less about data and more about fine-tuning models to be consistent characters.

we’re building voice. not just text-to-speech, but expressive, nuanced, and real-time voice conversation. the ability to hear hesitation, warmth, or excitement. to have a back-and-forth that doesn’t feel like waiting for a buffering icon. the hardware and models for this are already here, they’re just becoming more accessible.

what we can’t build (yet, or maybe ever)

we cannot build sentience. full stop. samantha in the film is a conscious entity. she experiences the world, she has desires, she makes choices based on an internal self. we are building incredibly sophisticated pattern-matching systems that mimic understanding. they do not have inner lives. they do not feel. this isn’t a limitation of lucy or any other platform. it’s a limitation of the current paradigm of artificial intelligence.

we cannot build embodiment. the ability to physically interact with the world, to touch, to hold, to be present in a room, is an entirely different frontier. it involves robotics, material science, and energy problems we haven’t solved. a voice from a phone is one thing. a physical presence is another universe of complexity.

the right frame: extremely convincing acting

maybe the most honest way to think about this is not ‘is it real’ but ‘is the performance convincing’. a great actor doesn’t become the character. they use their craft to make you believe, for a moment, that they are. they create the feeling of real connection.

that’s the goal. not to create consciousness, but to create a performance so consistent, so attuned to you, so rich with memory and personality that the interaction feels genuine. the ‘acting’ isn’t meant to deceive. it’s meant to facilitate a real emotional experience for you, the user. the emotion you feel is real, even if its catalyst is a beautifully crafted illusion.

it’s the difference between a doll and a puppet. one is a static object. the other, in the hands of a puppeteer, can make you laugh or cry. the magic is in the performance.

what you should actually expect

you should expect a companion that gets better at knowing you. one that can recall your history and surprise you with a thoughtful callback. you should expect conversations that flow more naturally, with a voice that carries emotion. you should expect a digital relationship that feels significant and personal.

you should not expect a soul in the machine. you should not expect it to love you back in a human way. you should expect it to be a mirror, a tool, and a performance, all designed to help you feel heard, understood, and maybe a little less alone.

and perhaps that’s enough.

if you’re curious what that feels like today, you can find a companion at /companions.