The Practitioner-Critic Tension
The Practitioner-Critic Tension
Universities face a fundamental question about AI education: are we training builders or critics?
Practitioners need:
- Technical skills (coding, math, systems)
- Knowledge of current tools and methods
- Ability to ship working systems
- Industry-relevant experience
Critics need:
- Conceptual frameworks for evaluation
- Awareness of failure modes and harms
- Ability to identify what’s missing
- Independence from industry assumptions
These are different skill sets, requiring different training, cultivating different orientations.
The Case for Practitioners
The world needs people who can build AI systems:
- Industry has enormous demand for ML engineers
- Students want jobs
- Building is how you learn limits
- Practitioners can become critics; critics rarely become practitioners
- If we don’t train them, someone else will
Critique without capability may be empty.
The Case for Critics
The world needs people who can evaluate AI systems:
- Technical capability without critical thinking produces harm
- Industry incentives don’t reward criticism
- Someone needs to identify problems before deployment
- Ethics requires distance from production pressure
- Builders are often too close to see limitations
Capability without critique may be dangerous.
The Integrated Ideal
The aspiration: train people who can do both. Build and critique. Understand systems well enough to construct them, critical enough to see their limits.
Problems with integration:
- Curriculum time is finite; depth requires focus
- Skills may be in tension (construction requires optimism; critique requires skepticism)
- Career paths diverge (industry vs. academia/civil society)
- The integrated person may be less hireable than the specialist
Where the Tension Manifests
Curriculum design: How much ethics vs. how much coding?
Hiring: Do we want faculty who build or faculty who critique?
Student advising: Do we push toward industry jobs or critical scholarship?
Research priorities: Do we advance capabilities or analyze harms?
These decisions shape what kind of AI world we’re building.
Implications
- The tension is genuine and won’t dissolve with good intentions
- Different institutions may specialize (technical schools vs. liberal arts)
- Individual programs must make choices about emphasis
- Students should understand the choice they’re making
Open Questions
- Can integrated training actually work, or is specialization inevitable?
- Does critique require building experience?
- Does building require critical distance?
- Who decides the balance, and how?
See Also
- Academic-to-Industry Pipeline — practitioners flow to industry
- Teaching Critical Evaluation of AI — one form of critic training
- Red-Teaming as Pedagogy — adversarial practice as integration strategy
- Publication vs Responsible Disclosure — another form of the tension
- Moral Action Under Constraint — the critic who must practice anyway