Uploaded October 2025 | Updated November 2026, 1 day ago
Ruru Hoong, Harvard Business School PhD candidate "Calibrated Coarsening: Designing Information for AI-Assisted Decisions"
Ruru Hoong, Harvard Business School PhD candidate "Calibrated Coarsening: Designing Information for AI-Assisted Decisions"

![Artificial Intelligence, Quantum and Beyond: 2026 MIT IDE Annual Conference
LLMs are better at short tasks than long ones — and that single pattern may tell you more about automation risk in your organization than any other metric. MIT IDEs AI Quantum and Beyond Research Group shares new frameworks and preliminary findings on how AI capabilities translate (or dont) into actual task loss.
In this conference session, researchers from MIT IDE and Future Tech present three interconnected perspectives on AI and work: how to map LLM capabilities to specific jobs, what AI risks matter most and to whom, how language models age in scientific use, and a new economic framework for thinking about partial versus full automation.
WHAT YOULL LEARN
— Task duration is a reliable proxy for AI automation risk: LLMs handle short tasks (~75% success rate) far better than tasks taking weeks (~40%), and that gap has real implications for which workers face the most disruption.
— Automation doesnt automatically hurt workers — what matters is whether the most expert or least expert parts of a job get automated, with opposite effects on wages and employment.
— Language model adoption in science follows a predictable inverted-U lifecycle, and that lifecycle is compressing — newer models peak faster and become obsolete sooner.
— Full automation is the exception, not the rule; partial automation is the dominant pattern, and its reach depends heavily on deployment scale, task complexity, and standardization.
CHAPTERS
0:00 Introduction
0:18 Neil Thompson: mapping LLM capabilities to tasks and workers
15:09 Peter Slattery: AI risk priorities and organizational response
29:18 Ana Trisovic: how science adopts language models
44:52 Martin Fleming: reconceptualizing automation along a spectrum
58:46 Q&A
FEATURED RESEARCHERS
Neil Thompson, Principal Research Scientist, MIT Sloan / MIT IDE / CSAIL
Research area: AI capabilities, technology evolution, workforce economics
Peter Slattery, [Title], MIT AI Risk Initiative
Research area: AI risk prioritization, organizational risk response
Ana Trisovic, Research Analyst, Future Tech
Research area: Language model adoption in science, AI bibliometrics
Martin Fleming, Research Scientist, MIT IDE / Future Tech; Former Chief Economist, IBM
Research area: Automation economics, LLM task analysis
ABOUT MIT IDE
The MIT Initiative on the Digital Economy (IDE) researches how AI, data, and digital technology reshape business, work, and society. Based at MIT Sloan School of Management.
Website: https://ide.mit.edu
#MITIDE #ArtificialIntelligence #FutureOfWork Artificial Intelligence, Quantum and Beyond: 2026 MIT IDE Annual Conference](https://i.ytimg.com/vi/msgVcIcIxE0/mqdefault.jpg)








