Iseminger AI Landscape Explainer Framework
Accurately map any AI-related technology, product, or concept onto the correct layer of the AI hierarchy — from broad AI down to Generative AI — so you can explain, evaluate, or position it with precision
// TL;DR
The Iseminger AI Landscape Explainer Framework is a structured method for classifying any AI technology onto the correct layer of a nested hierarchy: AI → Machine Learning → Deep Learning → Generative AI. Use it whenever you need to explain, compare, or position AI products and concepts for any audience — from boardrooms to classrooms. It provides an 8-step workflow that tests each technology against layer-specific characteristics, maps it on a historical timeline, identifies capabilities and abuse risks, and calibrates the explanation to the audience's technical level.
// When should you use the Iseminger AI Landscape Explainer Framework?
Use this skill whenever you need to explain, categorise, or compare AI technologies for an audience unfamiliar with the distinctions — or whenever a new AI tool, product, or trend needs to be placed in structural context relative to other AI concepts.
// What inputs do you need before applying the Iseminger framework?
- AI concept or technology to classifyrequired
The specific term, product, or technology you want to place in the hierarchy (e.g. ChatGPT, a fraud-detection system, a deepfake tool) - Audience technical levelrequired
How much prior knowledge the target audience has — beginner, intermediate, or expert — so simplifications can be calibrated appropriately - Use-case or domain context
The industry or scenario in which this AI technology is being applied (e.g. cybersecurity, entertainment, healthcare)
// What are the core principles behind the Iseminger AI Landscape Explainer Framework?
Nested Venn Diagram Structure
AI, Machine Learning, Deep Learning, and Generative AI are not separate silos — they are nested layers. Machine Learning is a subset of AI. Deep Learning is a subset of Machine Learning. Generative AI sits inside Deep Learning. Every classification decision must respect this containment hierarchy.
The Machine Is Learning (Not Programmed)
The defining characteristic of Machine Learning is that the system is not explicitly programmed with rules — it is given large amounts of data and it observes patterns. More training data yields higher prediction confidence. This distinction separates ML from earlier rule-based AI like expert systems.
Deep Means Multiple Layers
Deep Learning earns its name from having multiple layers of neural networks that simulate how the human brain operates. A key property is interpretive opacity: because there are so many layers, it can be difficult to fully decompose why a particular output was produced.
Foundation Models Drive the Adoption Explosion
Foundation Models — large-scale pre-trained models such as Large Language Models — are the architectural basis for Generative AI. They are what changed the AI adoption curve from a slow uptick to a near-vertical rise. Understanding any modern AI product requires identifying whether it is built on a Foundation Model.
Generative AI Creates New Content
Generative AI technologies produce new content — text, audio, video, images — rather than merely classifying or predicting from existing data. The music analogy applies: all notes already exist, yet new songs are genuinely created. Generation is the recombination of learned patterns into novel outputs, not mere regurgitation.
Simplification Is a Feature, Not a Bug
When explaining AI layers to non-experts, deliberate generalisation is required. Acknowledge upfront that simplifications are being made so that the audience can build a working mental model before diving into edge cases or exceptions.
// How do you apply the Iseminger AI Landscape Explainer Framework step by step?
- 1
Identify the broadest applicable layer
Ask: does this technology attempt to simulate something that would match or exceed human intelligence — learning, inference, reasoning? If yes, it belongs inside the outermost circle: Artificial Intelligence. Everything downstream is also AI, so start here and always confirm this baseline.
- 2
Test for Machine Learning characteristics
Ask: is the system learning from data rather than being explicitly programmed with rules? Is it spotting patterns, making predictions, or flagging outliers? If yes, classify it as Machine Learning (a subset of AI). Key signal: the more training data, the more confident its predictions.
- 3
Test for Deep Learning characteristics
Ask: does the system use neural networks — specifically multiple layers of them — to simulate brain-like processing? If yes, classify it as Deep Learning (a subset of ML). Note interpretive opacity as a defining property: outputs may be valid even when the internal reasoning is hard to decompose.
- 4
Test for Foundation Model and Generative AI characteristics
Ask: is the system built on a large pre-trained Foundation Model? Is it generating new content — text, audio, video, images? Examples include Large Language Models (LLMs), chatbots, and deepfake-generation tools. If yes, classify as Generative AI (innermost layer). Autocomplete is a useful on-ramp analogy: LLMs do not predict the next word — they predict the next sentence, paragraph, or full document.
- 5
Map the technology's timeline position
Contextualise the technology historically: Expert Systems (1980s–90s) → Machine Learning popularisation (2010s) → Deep Learning popularisation (2010s) → Foundation Models and Generative AI (recent explosion). This timeline explains why a technology feels mature or emergent and helps set audience expectations.
- 6
Identify the primary capability the technology delivers
Label the functional output: pattern recognition, prediction, outlier detection (classic ML); content generation — text, audio, video (Generative AI); summarisation of existing content (also Generative AI). Matching capability to layer prevents mislabelling.
- 7
Flag applicable use-cases and abuse risks
For each layer, name a legitimate use-case and, where relevant, an abuse vector. Generative AI in particular carries dual-use risk: chatbots and deepfakes both emerge from the same Foundation Model layer. Explicitly naming this prevents naive or dangerously optimistic framings.
- 8
Deliver the explanation calibrated to audience level
State simplifications openly. For non-technical audiences, use the autocomplete analogy for LLMs and the music analogy for generative content. For technical audiences, acknowledge that the nested Venn diagram is a deliberate generalisation, not a rigorous taxonomy.
// What are real-world examples of the Iseminger framework in action?
A product manager wants to explain to a board of directors why a fraud-detection product is 'AI' and how it differs from the company's new customer-service chatbot
Apply the Nested Venn Diagram: both products sit inside AI. The fraud-detection tool is Machine Learning — it was trained on transaction data, spots outliers (unusual user behaviour), and its confidence improves with more training data. The chatbot is Generative AI built on a Foundation Model (an LLM); it predicts and generates full conversational responses, not just the next word. Timeline context: the fraud tool uses 2010s-era ML maturity; the chatbot represents the recent adoption explosion driven by Foundation Models.
A journalist asks whether a viral audio clip that fakes a celebrity's voice is 'AI-generated'
Classify it through the layers: it is AI (simulates human capability) → Machine Learning (learned from training data) → Deep Learning (neural network layers) → Generative AI, specifically an audio Foundation Model. Label it a deepfake. Note the dual-use risk: the same technology has legitimate applications (e.g. preserving someone's voice when they lose the ability to speak) and clear abuse potential (fabricating statements a person never made).
A student asks whether the autocomplete on their phone is the same as ChatGPT
Use the autocomplete analogy as the bridge, then show the exponential leap: phone autocomplete predicts the next word. An LLM (a Foundation Model inside Generative AI) predicts the next sentence, paragraph, or entire document. Both use pattern recognition from training data (Machine Learning), but LLMs sit in the innermost Generative AI layer and produce genuinely new content — not just the most statistically likely next word.
// What mistakes should you avoid when using the Iseminger framework?
- Treating AI, Machine Learning, Deep Learning, and Generative AI as separate, parallel categories rather than nested layers — they are concentric, not coordinate
- Assuming Generative AI is 'just regurgitation' — new content is genuinely generated through recombination, analogous to how new music is composed from pre-existing notes
- Conflating the difficulty of interpreting Deep Learning outputs with the technology being unreliable — interpretive opacity is a structural property of multi-layer neural networks, not a defect
- Skipping the timeline contextualisation — without it, audiences cannot understand why adoption was slow for decades and then suddenly went 'straight to the moon'
- Over-explaining to non-expert audiences without first acknowledging that simplifications are being made — this erodes trust when experts push back on generalisations
- Treating deepfakes as purely a threat — failing to acknowledge legitimate entertainment, accessibility, and creative use-cases produces an unbalanced and therefore less credible explanation
// What key terms does the Iseminger AI Landscape Explainer Framework define?
- Artificial Intelligence (AI)
- The broad field of using computers to simulate capabilities that would match or exceed human intelligence — including learning, inference, and reasoning. The outermost layer of the Nested Venn Diagram.
- Machine Learning
- A subset of AI in which the machine learns by observing patterns in large amounts of training data rather than being explicitly programmed. Core capabilities: prediction and outlier detection. Confidence scales with training data volume.
- Deep Learning
- A subset of Machine Learning that uses multiple layers of neural networks to simulate brain-like processing. Named 'deep' because of the multiple layers. Characterised by interpretive opacity — outputs can be valid even when internal reasoning is hard to decompose.
- Neural Networks
- Computer structures that simulate and mimic the way the human brain operates, used as the architectural basis for Deep Learning.
- Interpretive Opacity
- The property of Deep Learning systems whereby the multi-layer structure makes it difficult to fully decompose and understand exactly why a particular output was produced.
- Generative AI (Gen AI)
- The innermost layer of the Nested Venn Diagram. AI technologies that generate new content — text, audio, video, images — built on Foundation Models. Includes chatbots, LLMs, and deepfake tools.
- Foundation Models
- Large-scale pre-trained models that serve as the architectural basis for Generative AI applications. Examples include Large Language Models. The emergence of Foundation Models is what changed the AI adoption curve from gradual to near-vertical.
- Large Language Models (LLMs)
- A type of Foundation Model that models language — predicting not just the next word (like autocomplete) but the next sentence, paragraph, or entire document. The basis for modern chatbots.
- Deepfakes
- Outputs of Generative AI audio or video Foundation Models that recreate a person's voice or likeness to produce content the person never actually created. Dual-use: legitimate applications in entertainment and accessibility; significant abuse potential.
- Nested Venn Diagram
- Iseminger's structural metaphor for the relationship between AI layers: AI contains ML, ML contains Deep Learning, Deep Learning contains Generative AI. Each inner layer inherits all properties of the outer layers.
- Expert Systems
- An earlier form of AI (prominent 1980s–90s) built on explicitly programmed rules using languages like Lisp or Prolog — contrasted with Machine Learning, which learns from data rather than being programmed.
- Adoption Explosion
- Iseminger's characterisation of how Foundation Models and Generative AI caused AI adoption to go 'straight to the moon' after decades of slow, gradual uptake.
// FREQUENTLY ASKED QUESTIONS
What is the Iseminger AI Landscape Explainer Framework?
It is a classification framework that places any AI technology onto the correct layer of a nested hierarchy: Artificial Intelligence contains Machine Learning, which contains Deep Learning, which contains Generative AI. Developed from David Iseminger's explainer approach, it provides an 8-step workflow to systematically test, classify, and explain AI concepts with precision calibrated to any audience's technical level.
What is the Nested Venn Diagram in AI?
The Nested Venn Diagram is the structural metaphor at the core of this framework. It shows that AI, Machine Learning, Deep Learning, and Generative AI are concentric layers — not separate parallel categories. Each inner layer inherits all properties of its outer layers. Machine Learning is a subset of AI, Deep Learning is a subset of ML, and Generative AI sits inside Deep Learning.
How do I classify an AI product using the Iseminger framework?
Start at the broadest layer and work inward. First confirm the technology qualifies as AI (simulates human intelligence). Then test whether it learns from data (Machine Learning), uses multi-layer neural networks (Deep Learning), and generates new content via a Foundation Model (Generative AI). Each step has specific diagnostic questions. Finally, map it on a timeline and calibrate your explanation to your audience.
How do I explain the difference between AI and machine learning to non-technical people?
AI is the broad umbrella — any technology that simulates human-level intelligence. Machine Learning is a specific approach within AI where the system learns patterns from large datasets instead of being explicitly programmed with rules. Use the nested circle visual: ML sits inside AI. Acknowledge upfront that you're simplifying, then use concrete examples like fraud detection (ML) versus a rule-based expert system (older AI).
How does the Iseminger framework compare to just Googling what type of AI something is?
A generic search often returns conflicting or siloed definitions that treat AI, ML, Deep Learning, and Gen AI as parallel categories. The Iseminger framework enforces the correct nested relationship and provides a repeatable 8-step workflow — including audience calibration, timeline positioning, and dual-use risk flagging — that a simple search cannot replicate. It produces structured, defensible explanations rather than fragmented definitions.
When should I use the Iseminger AI Landscape Explainer Framework?
Use it whenever you need to explain, categorize, or compare AI technologies for any audience — especially one unfamiliar with the distinctions between AI layers. It is particularly useful when positioning a new AI product, briefing executives, writing educational content, evaluating vendor claims, or placing a trending AI tool in structural context relative to other AI concepts.
Is ChatGPT machine learning or generative AI?
ChatGPT is both — and that is exactly the point of the nested hierarchy. It is AI (simulates human intelligence), Machine Learning (learns from data), Deep Learning (uses multi-layer neural networks), and Generative AI (generates new text content using a Foundation Model, specifically a Large Language Model). It sits in the innermost layer but inherits every property of all outer layers.
What results can I expect from using the Iseminger framework?
You will produce clear, structurally accurate explanations that position any AI technology on the correct layer of the hierarchy. Audiences gain a working mental model of how AI concepts relate to each other. Executives can make better-informed decisions, educators deliver more precise lessons, and marketers avoid mislabeling products. The framework also surfaces dual-use risks, preventing naively optimistic or fear-driven narratives.
What's the difference between deep learning and machine learning?
Deep Learning is a subset of Machine Learning. All Deep Learning is ML, but not all ML is Deep Learning. The distinguishing feature is architecture: Deep Learning uses multiple layers of neural networks that simulate brain-like processing. This multi-layer structure gives Deep Learning its name and also produces interpretive opacity — valid outputs whose internal reasoning can be difficult to fully decompose.
Why did AI adoption suddenly explode after decades of slow growth?
Foundation Models — large-scale pre-trained models like Large Language Models — are what changed the adoption curve from a slow uptick to near-vertical growth. The Iseminger framework calls this the 'Adoption Explosion.' Foundation Models enabled Generative AI applications that produce text, images, audio, and video, making AI tangible and useful for mainstream audiences for the first time.