We Thought We Were Programming AI — We Were Just Talking to It

2026-05-31 · Source: Artificial Intelligence on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Intermediate, quick

Summary

The article explores the evolving perception of interacting with AI, particularly Large Language Models. Initially, prompt engineering was seen as a novel programming paradigm, where refining natural language instructions and examples yielded predictable, deterministic outputs. This approach made language the primary interface, allowing users to "debug" by rewriting sentences. However, this illusion of control gradually eroded as prompts became inconsistent, and systems struggled with variations not explicitly covered. This shift suggests that interacting with AI is less about deterministic programming and more akin to a conversational, less predictable engagement with an intelligent agent.

Key takeaway

For Prompt Engineers developing AI applications, recognize that your interaction with LLMs is shifting from deterministic programming to a more conversational, less predictable engagement. You should anticipate inconsistencies and move beyond simple prompt refinement, exploring strategies that account for the AI's evolving and less stable behavior. This necessitates a deeper understanding of model nuances rather than relying solely on linguistic adjustments.

Key insights

The initial view of prompt engineering as deterministic programming is giving way to understanding AI interaction as less predictable conversation.

Principles

• Language serves as the AI interface.
• AI interaction is not deterministic.

In practice

• Refine prompts with examples.
• Adjust wording and constraints.

Topics

• Prompt Engineering
• Large Language Models
• AI Interaction
• System Determinism
• Conversational AI

Best for: NLP Engineer, AI Product Manager, Prompt Engineer, AI Engineer, Machine Learning Engineer

Related on AIssential

• Prompt Engineering: Saying what matters most — Effective prompts are critical for guiding LLMs, defining behavior, controlling output quality, and managing operational costs.
• Architecting Production-Grade Agents through LLM Orchestration and Agentic Loops — Reliable AI systems require deterministic agentic architecture over prompt-driven LLM experimentation.
• Code, Chaos, and the Invisible Logic of LLMs — LLMs predict words probabilistically, not with human intention, leading to potential misunderstandings in code generation.
• App lets you put multiple LLMs into one group chat! Have them fight it out or collaborate. — Interacting LLMs in group chats can reveal novel insights and emergent conversational behaviors.
• Your AI Agent Isn’t Dumb. It Has ADHD — LLMs forget instructions mid-conversation due to context window attention bias, causing significant performance drops.
• stop lying to your ai — AI requires brutally honest, explicit communication to deliver desired outputs, unlike human interaction.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence on Medium.