Publications
Publications
Show Results For
- All HBS Web
(9,623)
- Faculty Publications (2,034)
- 2023
- Article
MoPe: Model Perturbation-based Privacy Attacks on Language Models
By: Marvin Li, Jason Wang, Jeffrey Wang and Seth Neel
Recent work has shown that Large Language Models (LLMs) can unintentionally leak sensitive information present in their training data. In this paper, we present Model Perturbations (MoPe), a new method to identify with high confidence if a given text is in the training... View Details
- 2023
- Article
Post Hoc Explanations of Language Models Can Improve Language Models
By: Satyapriya Krishna, Jiaqi Ma, Dylan Slack, Asma Ghandeharioun, Sameer Singh and Himabindu Lakkaraju
Large Language Models (LLMs) have demonstrated remarkable capabilities in performing complex tasks. Moreover, recent research has shown that incorporating human-annotated rationales (e.g., Chain-of-Thought prompting) during in-context learning can significantly enhance... View Details
- December 2023
- Article
Self-Orienting in Human and Machine Learning
By: Julian De Freitas, Ahmet Uğuralp, Zeliha Uğuralp, Laurie Paul, Joshua B. Tenenbaum and T. Ullman
A current proposal for a computational notion of self is a representation of one’s body in a specific time and place, which includes the recognition of that representation as the agent. This turns self-representation into a process of self-orientation, a challenging... View Details
- 2023
- Other Article
The Harvard USPTO Patent Dataset: A Large-Scale, Well-Structured, and Multi-Purpose Corpus of Patent Applications
By: Mirac Suzgun, Luke Melas-Kyriazi, Suproteem K. Sarkar, Scott Duke Kominers and Stuart Shieber
Innovation is a major driver of economic and social development, and information about many kinds of innovation is embedded in semi-structured data from patents and patent applications. Though the impact and novelty of innovations expressed in patent data are difficult... View Details
- 2023
- Article
Verifiable Feature Attributions: A Bridge between Post Hoc Explainability and Inherent Interpretability
By: Usha Bhalla, Suraj Srinivas and Himabindu Lakkaraju
With the increased deployment of machine learning models in various real-world applications, researchers and practitioners alike have emphasized the need for explanations of model behaviour. To this end, two broad strategies have been outlined in prior literature to... View Details
- December 2023
- Article
When Should the Off-Grid Sun Shine at Night? Optimum Renewable Generation and Energy Storage Investments
By: Christian Kaps, Simone Marinesi and Serguei Netessine
Globally, 1.5 billion people live off the grid, their only access to electricity often limited to operationally-expensive fossil fuel generators. Solar power has risen as a sustainable and less costly option, but its generation is variable during the day and... View Details
- November 2023
- Case
Open Source Machine Learning at Google
Set in early 2023, the case exposes students to the challenges of managing open source software at Google. The case focuses on the challenges for Alex Spinelli, Vice President of Product Management for Core Machine Learning. He must set priorities for Google’s efforts... View Details
- November 2023
- Case
From Imitation to Innovation: Zongshen Industrial Group (Abridged)
By: Willy Shih and Nancy Dai
Like other small shops based in Chongqing, China, Zongshen Industrial Group started by assembling motorcycles from "standard" parts. The quality of its early products was good enough for rural Chinese buyers, though wealthier consumers usually purchased premium... View Details
- November 2023 (Revised July 2024)
- Case
Aviva plc: Examining Net Zero
The board of Aviva Plc, one of the world’s largest insurers, must review its climate risk exposures and evaluate next steps. Risk experts at the firm have conducted a robust set of analyses prepared for its regulator, the Bank of England, simulating how various climate... View Details
- November 2023
- Case
Copilot(s): Generative AI at Microsoft and GitHub
This case tells the story of Microsoft’s 2018 acquisition of GitHub and the subsequent launch of GitHub Copilot, a tool that uses generative artificial intelligence to suggest snippets of code to software developers in real time. Set in late 2021, when Copilot was... View Details
- 2023
- Working Paper
Learning by Investing: Entrepreneurial Spillovers from Venture Capital
By: Josh Lerner, Jinlin Li and Tong Liu
This paper studies how investing in venture capital (VC) affects the entrepreneurial outcomes of individual limited partners (LPs). Using comprehensive administrative data on entrepreneurial activities and VC fundraising and investments in China, we first document that... View Details
- November 2023 (Revised February 2025)
- Background Note
Corporate Climate Targets
By: Willy C. Shih, Michael W. Toffel and Kelsey Carter
Companies that are addressing climate change by mitigating their greenhouse gas emissions often set reduction targets. This note describes several types of widely used carbon reduction targets, including carbon neutral, science based, net zero, real zero, and carbon... View Details
- 2024
- Working Paper
What Do Impact Investors Do Differently?
In recent years, impact investors – private investors who seek to generate simultaneously financial and social returns – have attracted intense interest and controversy. We analyze a novel, comprehensive data set of impact and traditional investors to assess how the... View Details
- October 2023 (Revised February 2024)
- Technical Note
Design and Evaluation of Targeted Interventions
By: Eva Ascarza and Ta-Wei (David) Huang
Targeted interventions serve as a pivotal tool in business strategy, streamlining decisions for enhanced efficiency and effectiveness. This note delves into two central facets of such interventions: first, the design of potent decision guidelines, or targeting... View Details
- October 2023
- Case
Taiwan After Globalization: Twilight of the Developmental State?
By: Debora L. Spar and Julia M. Comeau
In the last 70 years, the small island of Taiwan has achieved what many believe to be a “miracle”: its economy has grown at a record-setting pace, driven and guided by one of the world's most successful set of industrial policies, and it has become one of the richest... View Details
- 2023
- Working Paper
Are Hospital Quality Indicators Causal?
By: Amitabh Chandra, Maurice Dalton and Douglas O. Staiger
Hospitals play a key role in patient outcomes and spending, but efforts to improve their quality are hindered because we do not know whether hospital quality indicators are causal or biased. We evaluate the validity of commonly used quality indicators, such as... View Details
- 2023
- Working Paper
Black-box Training Data Identification in GANs via Detector Networks
By: Lukman Olagoke, Salil Vadhan and Seth Neel
Since their inception Generative Adversarial Networks (GANs) have been popular generative models across images, audio, video, and tabular data. In this paper we study whether given access to a trained GAN, as well as fresh samples from the underlying distribution, if... View Details
- 2023
- Working Paper
Causal Interpretation of Structural IV Estimands
By: Isaiah Andrews, Nano Barahona, Matthew Gentzkow, Ashesh Rambachan and Jesse M. Shapiro
We study the causal interpretation of instrumental variables (IV) estimands of nonlinear, multivariate structural models with respect to rich forms of model misspecification. We focus on guaranteeing that the researcher's estimator is sharp zero consistent, meaning... View Details
Keywords: Mathematical Methods
- 2023
- Working Paper
In-Context Unlearning: Language Models as Few Shot Unlearners
By: Martin Pawelczyk, Seth Neel and Himabindu Lakkaraju
Machine unlearning, the study of efficiently removing the impact of specific training points on the
trained model, has garnered increased attention of late, driven by the need to comply with privacy
regulations like the Right to be Forgotten. Although unlearning is... View Details