00 — Insights & resources
Resources
Plain-language references for professionals and L&D teams.
Last reviewed: October 2026
01 — Glossary
Glossary of key terms
The vocabulary used in our workshops, from the institute’s own terminology to the technical terms behind current tools.
- Super Intelligence (SI)
- The term SI Institute uses for today’s intelligent technologies — language models, assistants, agents and related systems. It is a deliberate naming choice, not a claim that artificial superintelligence exists. Elsewhere, these technologies are usually called artificial intelligence (AI).
- Artificial intelligence (AI)
- The broad field of computer systems that perform tasks associated with human intelligence, such as understanding language, recognizing patterns and making predictions.
- Artificial superintelligence
- A hypothetical system that would exceed human intelligence across virtually all domains. No such system exists today, and our programs make no such claim.
- Large language model (LLM)
- A model trained on very large amounts of text to predict and generate language. LLMs power most current assistants and copilots.
- Multimodal model
- A model that works with several types of input and output, such as text, images, audio, documents and video.
- Reasoning model
- A model that spends additional computation working through a problem step by step before answering. Useful for complex analysis; slower and more costly for simple tasks.
- Assistant
- A conversational application, such as ChatGPT, Claude or Gemini, that responds to instructions and can work with files, search and tools.
- Copilot
- An assistant embedded in a work application — email, documents, spreadsheets, code editors — that can use the content of that application within the user’s permissions.
- Agent
- An SI system that plans steps, uses tools and takes actions to reach a goal within defined limits, often with human approval at key points.
- Multi-agent orchestration
- Several agents with different roles coordinated to complete a larger task. Promising, but still largely emerging or experimental for business-critical processes.
- Prompt engineering
- Designing the instructions given to a model — goal, context, constraints, examples and output format — to get reliable results.
- Context engineering
- Selecting, structuring and supplying the information a model needs for a task (documents, data, instructions, history), and leaving out what distracts it.
- Context window
- The amount of text and other input a model can take into account at once. Larger windows allow longer documents but do not guarantee that every detail is used.
- Retrieval-augmented generation (RAG)
- A method in which the system first retrieves relevant passages from a document collection and then generates an answer based on them, ideally with citations.
- Hallucination
- Output that sounds plausible but is false or unsupported, such as invented facts, figures or sources.
- Grounding
- Tying model output to provided sources or data, so that answers can be checked against them.
- Fine-tuning
- Further training a model on specific examples to adapt its behavior. Most business use cases are better served by good instructions and retrieval first.
- Model Context Protocol (MCP)
- An open standard for connecting SI applications and agents to tools and data sources in a consistent way.
- Evaluation
- Systematically testing SI output against defined criteria and test cases, to measure quality before and after changes.
- Human-in-the-loop
- A design in which people review, approve or correct SI output at defined points, especially before consequential actions.
- Prompt injection
- An attack in which hidden or malicious instructions in content (a web page, email or document) try to make an SI system act against its intended rules.
- AI literacy
- The skills and understanding needed to use AI systems appropriately. The EU AI Act requires providers and deployers of AI systems to take measures to ensure sufficient AI literacy among their staff.
02 — Maturity map
Established, emerging, experimental
What intelligent technology can reliably do today, what is maturing, and what is not yet ready for unsupervised business use — and what each means for training. Reviewed with every curriculum update.
Established
Widely deployed and used in daily work by many organizations.
- Drafting, summarizing, rewriting and translating textTrain every employee; focus on quality checks and responsible use.
- Analyzing documents and answering questions about themTeach source discipline and verification of citations.
- Copilots inside email, documents, meetings and spreadsheetsTrain on concrete workflows per role; adoption depends on habits.
- Data analysis with code-executing assistantsTeach result validation; keep humans accountable for numbers.
- Coding assistance in the editorCombine with secure code review and team conventions.
- Knowledge assistants over curated document sets (RAG)Teach content preparation, evaluation and maintenance.
Emerging
Used in production for bounded tasks; practices are still maturing quickly.
- Deep-research features that search and synthesize many sourcesUseful starting points; every claim still needs verification.
- Agents using tools for bounded business tasksPilot with clear permissions, approval steps and monitoring.
- Coding agents completing multi-file changesDelegate well-specified tasks; review every change.
- Agents operating browsers and desktop applicationsExperiment in sandboxes; not yet for unsupervised production use.
Experimental
Promising, but not yet reliable enough for unsupervised business use.
- Autonomous multi-agent systems running whole processes end to endFollow and experiment; do not plan core operations around it yet.
- Agents taking consequential actions without human approvalKeep humans in the loop for decisions with real-world impact.
Artificial superintelligence — systems exceeding human intelligence across virtually all domains — does not exist and is not part of this map. “Super Intelligence (SI)” is our name for the technologies listed here.
03 — Planning guide
Planning SI learning in your annual training plan
Start from business priorities
Link SI learning to the outcomes the organization is pursuing this year — productivity, customer experience, risk, growth — rather than to the technology itself.
Segment your populations
All employees, people managers, senior leaders, functional specialists and technical teams need different depths. One course for everyone rarely works.
Map to your competency framework
Place SI capabilities inside your existing technical, behavioral, leadership, management and functional competency families instead of creating a separate silo.
Use the programs you already run
Leadership, HiPo, graduate and management programs and functional academies are natural places for SI modules — with cohorts, sponsors and budgets already in place.
Choose formats per population
Live workshops for skills that need practice; digital learning for reach and reinforcement; blended journeys where daily habits must change.
Measure application, not attendance
Track what participants do differently after the training: workflows adopted, outputs produced, time saved, quality improved.
04 — Checklist
Questions to ask any SI training provider
Including us. A short checklist for L&D teams comparing training offers.
- 01Does each program state what participants will be able to do afterwards?
- 02How much of the time is hands-on practice on realistic work?
- 03Are the tools used the ones your organization has approved — or can they be?
- 04Is the content current, and how often is it reviewed?
- 05Does the provider distinguish established capabilities from experimental ones?
- 06Is responsible use, including data protection and verification, built into every program?
- 07Can the program connect to your competency framework and talent programs?
- 08Is the provider transparent about what it delivers itself and what depends on partners?
05 — Contact
Plan your next step
- For organizationsRequest a corporate proposalTell us about your audience, objectives and preferred format. We reply with questions or a first program outline.
- For professionalsRegister interest in a workshopBe informed as soon as public dates are confirmed for the workshops you are interested in.
Partnerships, media or other questions? Send a general inquiry.