Ted Johnson argues that current AI interactions rely on an outdated, batch-processing “prompt” system akin to punch cards, which forces users to adapt to machine limitations rather than enabling natural, conversational communication. He advocates for evolving AI interfaces to dynamically engage in dialogue by adapting to human communication styles, thereby reducing cognitive load and unlocking AI’s full potential as an intuitive, collaborative partner.
In his talk, Ted Johnson challenges the familiar experience of interacting with AI through prompts, likening the current method to using punch cards—a batch processing system from early computing. Despite the incredible advances in AI’s ability to understand natural language, the interface we use to communicate with it remains largely unchanged: a simple text box where users must package their entire request, submit it, and wait for a response. This outdated protocol forces users to learn “prompt engineering,” a skill akin to mastering punch card programming, rather than enabling a natural, conversational interaction with AI.
Johnson introduces three key concepts to frame his argument: the channel, expression, and protocol. The channel is the medium through which we communicate with machines, such as keyboards or microphones. Expression refers to the richness and range of meaning that can be conveyed through that channel. While expression has dramatically expanded with AI’s ability to understand natural language, the protocol—the rules governing the interaction—has not evolved and remains a batch process. This mismatch creates friction, making AI feel unnatural and difficult to use despite its power.
He illustrates the limitations of the current protocol with examples from voice-based AI systems that cannot distinguish when speech is directed at them or when to take turns in conversation. Emerging research models like Nvidia’s Personal Plex demonstrate more advanced conversational abilities, such as real-time turn-taking and active listening cues, showing promise for AI interfaces that participate dynamically in dialogue rather than waiting passively for complete inputs. These developments highlight the potential for AI to engage more naturally and collaboratively with users.
Johnson argues that AI should not just be seen as a smarter machine but as a new kind of interface technology that can remove traditional constraints and amplify human potential. Instead of forcing humans to adapt to machine limitations, AI interfaces should adapt to human communication styles, managing timing, modality, and context to reduce the cognitive load on users. This shift would enable more intuitive, human-centered interactions where AI understands and participates in conversations fluidly, rather than requiring precise, pre-packaged commands.
Ultimately, Johnson calls for a reimagining of AI interfaces that moves beyond the batch-processing prompt model to embrace the full richness of human communication. By doing so, we can eliminate the “translation tax” imposed by current systems and create AI that truly understands us without requiring us to reshape our natural ways of expressing intent. This evolution in interface design is crucial for broader adoption and for realizing AI’s potential as a partner in human creativity and problem-solving.