Machine learning increasingly intertwines technical problems with human and social questions
Evidence
How to prevent such a catastrophic divergence—how to ensure that these models capture our norms and values, understand what we mean or intend, and, above all, do what we want—has emerged as one of the most central and most urgent scientific questions in the field of computer science. It has a name: the alignment problem. — Brian Christian, The Alignment Problem, 65 (
#expands, via linked claim)
Categorization
- corroborated: 2 independent passages support this claim
- XREF: connects to: AI researchers aim to align general-purpose systems with hard-to-specify human intentions; Word embeddings serve as novel objects for sociological analysis of bias; Machine learning systems are not inherently fair or just by default (+20 more)
Related
- Ensuring AI captures human intent is now a central computer science problem (similarity 0.85)
- AI researchers aim to align general-purpose systems with hard-to-specify human intentions (similarity 0.84)
- Brian Christian — The Alignment Problem (similarity 0.81)
- Hundreds of papers missed blatant sexism in word embeddings (similarity 0.80)
- AI safety concerns have shifted from fringe to central research focus (similarity 0.80)
- Machine learning systems are not inherently fair or just by default (similarity 0.80)
- Machines could learn human values by observing behavior instead of being programmed (similarity 0.80)
- Word embeddings serve as novel objects for sociological analysis of bias (similarity 0.80)
- AI alignment may require machines to pursue human objectives, not their own (similarity 0.80)
- Copyright’s chilling effect on AI training data worsens algorithmic bias (similarity 0.79)
- Intent-alignment doesn’t specify whose intentions matter, leaving trade-off conflicts unresolved (similarity 0.79)
Library Connections
- Machine learning systems are not inherently fair or just by default
#exemplifies(0.85) - Even massive, comprehensive text datasets reflect biases present in the real world
#exemplifies(0.80) - Machine-learning bias often stems from unrepresentative training data
#exemplifies(0.80) - Large-scale English text corpora reflect real-world societal biases
#exemplifies(0.80)
Broader Connections
- Ensuring AI captures human intent is now a central computer science problem
#expands(0.85) — Brian Christian, The Alignment Problem, 65 - Word embeddings’ gender skew correlates with real occupational gender ratios
#exemplifies(0.78) - AI safety concerns have shifted from fringe to central research focus
#corroborates(0.75) - AI interactions can influence human emotional responses and behaviors.
#exemplifies(0.70) — Joanna Stern, I Am Not a Robot- My Year Using AI to…, 583 - Rule-based machine ethics produce unintended emergent consequences
#exemplifies(0.70) — , The Braided Reading EPUB, 1591 - AI integration raises crucial questions about human creativity.
#exemplifies(0.60) — Joanna Stern, I Am Not a Robot- My Year Using AI to…, 26 - AI advancements challenge society’s trust in online interactions.
#exemplifies(0.60) — Joanna Stern, I Am Not a Robot- My Year Using AI to…, 1175 - AI is influencing parenting and educational practices.
#exemplifies(0.60) — Joanna Stern, I Am Not a Robot- My Year Using AI to…, 1247 - The integration of AI in education may profoundly alter student experiences.
#exemplifies(0.60) — Joanna Stern, I Am Not a Robot- My Year Using AI to…, 1037 - Automation requires active human work to organize the relinquishing of subjectivity
#exemplifies(0.60) — Grégoire Chamayou, A Theory of the Drone, 650 - The effects of AI on emotional well-being can reflect broader societal trends.
#expands(0.55) — Joanna Stern, I Am Not a Robot- My Year Using AI to…, 1160