Skip to content
AImpact
IT EN
← Reading paths
◦

Reading path

Prompt Engineer

Master advanced prompting techniques and injection defense.

For those who want to go beyond basic prompts: chain-of-thought, few-shot learning, system prompt design, and prompt injection defense. This path covers the key milestones that redefined how we communicate with and control the behavior of language models.

  1. 01

    Why it matters to you

    InstructGPT proved that RLHF fundamentally changes how a model responds to instructions — the theoretical backbone of any effective system prompt.

    High Foundation Models

    InstructGPT: the fine-tuning that teaches GPT to obey

    OpenAI introduces InstructGPT: a GPT-3 refined with human feedback (RLHF) that follows instructions better than the 175B base model despite being much smaller (1.3B parameters).
  2. 02

    Why it matters to you

    Constitutional AI introduced explicit behavioral rules for models, a concept directly applicable when building robust and safe system prompts.

    Medium AI Security

    Constitutional AI: the model self-corrects without humans in the loop

    Anthropic publishes Constitutional AI: instead of pure RLHF, the model critiques and revises its own responses following a written 'constitution'. Less human labeling, more transparency.
  3. 03

    Why it matters to you

    GPT-4 dramatically improved instruction-following for complex, multi-step prompts, making advanced techniques like structured XML prompting and chain-of-thought reliable.

    Landmark Foundation Models

    GPT-4: the reasoning leap that resets the baseline

    OpenAI releases GPT-4, multimodal (text + image), with reasoning, coding, and reliability clearly beyond GPT-3.5. Passes bar, medical, and coding exams.
  4. 04

    Why it matters to you

    Function calling turns a prompt from free text into a structured interface — understanding its mechanics is essential for designing prompts that safely orchestrate tools.

    High AI Infrastructure

    Function calling: GPT learns to speak JSON

    OpenAI adds 'function calling' to the API: the model returns structured JSON conforming to a schema, enabling reliable tool integrations without fragile prompt engineering.
  5. 05

    Why it matters to you

    Reasoning models shift the prompting paradigm: less explicit chain-of-thought scaffolding, more precise problem framing — a fundamental shift every prompt engineer must understand.

    Landmark Foundation Models

    o1: the first model that 'thinks before answering'

    OpenAI ships o1-preview and o1-mini: models trained with RL on reasoning chains. On math, physics, competitive coding they beat GPT-4o by a huge margin. Paradigm shift.
  6. 06

    Why it matters to you

    The MCP spec standardizes how models interact with external tools, giving prompt engineers a protocol-level foundation for designing secure, controllable agents.

    High AI Infrastructure

    Model Context Protocol: the open standard to connect LLMs and data

    Anthropic open-sources the Model Context Protocol (MCP), a JSON-RPC standard that lets AI assistants talk to tools, file systems, databases, and SaaS without per-model ad-hoc integrations.
  7. 07

    Why it matters to you

    Image prompting introduced concepts like style weights, negative prompts, and semantic composition that later influenced advanced text prompting techniques.

    Landmark Image & Video Gen

    Stable Diffusion: image generation goes open

    Stability AI publicly releases weights and code of a text-to-image latent diffusion model that runs on a consumer GPU. AI image generation leaves the cloud.