"When David Marr at MIT moved into computer vision, he generated a lot of excitement, but he hit up against the problem of knowledge representation; he had no good representations for knowledge in his vision systems"
About this Quote
Minsky’s line has the clipped generosity of a rival who respects the brilliance but won’t let the legend stand unexamined. David Marr is introduced as the spark - the MIT wunderkind who made computer vision feel inevitable - and then Minsky pivots to the brick wall: representation. The praise is real, but it’s also a setup for Minsky’s larger point about AI’s chronic blind spot. Vision isn’t just optics plus clever math; it’s a constant negotiation with what the system already "knows" about the world.
The subtext is an institutional argument disguised as a technical critique. Marr’s program leaned hard into elegant computational frameworks and layered processing; Minsky is reminding the listener that elegance collapses without a vocabulary for everyday concepts: objects, intentions, categories, context, exceptions. In other words, the scene understanding that humans do without noticing. By saying Marr had "no good representations", Minsky isn’t nitpicking implementation details. He’s pointing to a missing bridge between perception and cognition - the gap that turns a camera into a sensor, not an agent.
Context matters: this is the late-20th-century AI civil war between approaches that trusted geometry and signal processing and those that obsessed over symbols, frames, and commonsense reasoning. Minsky’s jab lands because it’s still recognizably true: many breakthroughs in vision thrill the field, then stumble on the messier question of meaning. The quote is less eulogy than warning label: excitement is cheap; knowledge is the hard part.
The subtext is an institutional argument disguised as a technical critique. Marr’s program leaned hard into elegant computational frameworks and layered processing; Minsky is reminding the listener that elegance collapses without a vocabulary for everyday concepts: objects, intentions, categories, context, exceptions. In other words, the scene understanding that humans do without noticing. By saying Marr had "no good representations", Minsky isn’t nitpicking implementation details. He’s pointing to a missing bridge between perception and cognition - the gap that turns a camera into a sensor, not an agent.
Context matters: this is the late-20th-century AI civil war between approaches that trusted geometry and signal processing and those that obsessed over symbols, frames, and commonsense reasoning. Minsky’s jab lands because it’s still recognizably true: many breakthroughs in vision thrill the field, then stumble on the messier question of meaning. The quote is less eulogy than warning label: excitement is cheap; knowledge is the hard part.
Quote Details
| Topic | Artificial Intelligence |
|---|
More Quotes by Marvin
Add to List
