Philosophy Eats AI — Synthesis for MSBAi
Source: Schrage & Kiron, MIT Sloan Management Review, January 2025
Original: sloanreview.mit.edu/article/philosophy-eats-ai
Ingested: 2026-04-19 (PDF in discussions/raw/attachments/2026-04-19-ingest-article-1145/)
Core Argument
Ethics is just a small slice of philosophy’s role in AI. Three deeper frameworks shape whether AI creates or destroys value:
- Teleology — what is the AI for? What outcomes should it pursue?
- Epistemology — what counts as knowledge? How should uncertainty be expressed?
- Ontology — how does the AI represent reality? What entities and relationships does it recognize?
The Google Gemini failure wasn’t an ethics bug — it was a teleological failure: competing objectives (accuracy vs. DEI) were never reconciled, so the model optimized for the wrong one.
“AI systems rise or fall to the level of their philosophical training, not their technical capabilities.”
“Pattern matchers without purpose, computers that generate the wrong answers faster.”
The Three Frameworks for Agentic AI
1. Epistemological Agency
- Self-directed learning: identifies knowledge gaps proactively
- Dynamic hypothesis testing: generates and tests possibilities, not just pattern-matches
- Meta-cognitive awareness: communicates confidence levels and knowledge gaps
2. Ontological Understanding
- Self-understanding: maintains awareness of own capabilities and limits
- Causal architecture: builds models of how elements influence each other
- Systems thinking: recognizes nested complexity
3. Teleological Architecture
- Forms and refines goals autonomously
- Navigates purpose hierarchies: immediate actions → broader mission
- Resolves competing priorities with principled trade-offs
4. Ethical Agency Beyond rule-following: autonomous moral reasoning in novel situations. Transparent, interpretable, explainable choices.
MSBAi Implications
Curriculum connections
| Article concept | MSBAi touchpoint |
|---|---|
| Teleological clarity | Our AI-native operating model needs an explicit purpose statement — not just “AI in every course” but why (what does the AI help students accomplish that they couldn’t otherwise?) |
| Epistemological training | Cognitive offloading U-curve (PARKED-004): Zone 2 is precisely “epistemologically untrained” AI use — students offload without understanding |
| Ontological grounding | BADM 554 Week 1: before students use AI tools, they need a mental model of what the AI represents and what it misses |
| Teleological chaos (Gemini failure) | Perfect case study for BADM 557 (BI/strategy) or BADM 576 (LLMOps) |
| “Choice architect” framing | Our AIAS level design is a choice architecture — we scaffold when AI is available and when it isn’t, deliberately shaping student reasoning |
| Outputs → Outcomes | Directly maps to our assessment philosophy: no exams, real projects, oral defense. We teach to outcomes, not outputs |
| “Pattern matchers without purpose” | Exactly the risk in AIAS Level 1 use — students who submit AI-generated work without engaging with it are pattern-matchers |
For ai_native_strategy.md
The article gives us language to sharpen the AI-native, human-centered positioning:
The MSBAi program doesn’t just use AI — it trains students to reason about AI: what it’s for, what it knows, and how it represents the problems it’s solving. That’s the difference between producing AI operators and AI strategists.
The Schrage/Kiron framework (teleology + epistemology + ontology) could anchor the “critical AI engagement” positioning — distinguishing MSBAi from programs that teach AI as a toolset.
For design_principles.md
Principle 2 (Critical Engagement with AI) already cites Hardman (2026) on cognitive offloading. This article adds a strategic argument for why philosophical grounding matters — not just pedagogically but for organizational AI value creation. Students who graduate with epistemological/ontological clarity about AI are more valuable to employers than those who can only operate tools.
For Agentic AI course
The philosophical frameworks for agentic AI (Section 3 of the article) are essentially a reading list for the course. The supply chain scenario (pattern-matching response vs. philosophically-trained response) is a strong in-class exercise.
The Ocasio connection
Professor Ocasio’s 2am exchange applied exactly these frameworks to K-ai:
- Ontological question (Q6): what is the “organizational memory” really? Is “memory” the right word?
- Epistemological question (Q7): how is knowledge structured? does it scale?
- Teleological question (Q8): can you build a taxonomy? what’s the system’s purpose?
He was doing live philosophical audit on the system.
Parked for deeper integration
See PARKED-005 (to be added): Schrage/Kiron philosophical framework as curriculum spine for Agentic AI course + sharpening of MSBAi AI-native, human-centered positioning language.
Key quotes for use
“Acting as if philosophy and philosophical insights are incidental or incremental to enterprise AI impact minimizes their potential technological and economic impact.”
“Every prompt, parameter, and deployment encodes philosophical assumptions about knowledge, truth, purpose, and value.”
“The map is not the territory.” — Korzybski (cited in article on metrics vs. reality)
“We don’t rise to the level of our expectations; we fall to the level of our training.” — Archilochus (cited re: agentic AI)