Representative Bodies in the Age of AI

Curated on Posted on by Stefaan Verhulst

Report by POPVOX: “The report tracks current developments in the U.S. Congress and internationally, while assessing the prospects for future innovations. The report also serves as a primer for those in Congress on AI technologies and methods in an effort to promote responsible use and adoption. POPVOX endorses a considered, step-wise strategy for AI experimentation, underscoring the importance of capacity building, data stewardship, ethical frameworks, and insights gleaned from global precedents of AI in parliamentary functions. This ensures AI solutions are crafted with human discernment and supervision at their core.

Legislatures worldwide are progressively embracing AI tools such as machine learning, natural language processing, and computer vision to refine the precision, efficiency, and, to a small extent, the participatory aspects of their operations. The advent of generative AI platforms, such as ChatGPT, which excel in interpreting and organizing textual data, marks a transformative shift for the legislative process, inherently a task of converting rules into language.

While nations such as Brazil, India, Italy, and Estonia lead with applications ranging from the transcription and translation of parliamentary proceedings to enhanced bill drafting and sophisticated legislative record searches, the U.S. Congress is prudently venturing into the realm of Generative AI. The House and Senate have initiated AI working groups and secured licenses for platforms like ChatGPT. They have also issued guidance on responsible use…(More)”.

Ground Truths Are Human Constructions

Curated on Posted on by Stefaan Verhulst

Article by Florian Jaton: “Artificial intelligence algorithms are human-made, cultural constructs, something I saw first-hand as a scholar and technician embedded with AI teams for 30 months. Among the many concrete practices and materials these algorithms need in order to come into existence are sets of numerical values that enable machine learning. These referential repositories are often called “ground truths,” and when computer scientists construct or use these datasets to design new algorithms and attest to their efficiency, the process is called “ground-truthing.”

Understanding how ground-truthing works can reveal inherent limitations of algorithms—how they enable the spread of false information, pass biased judgments, or otherwise erode society’s agency—and this could also catalyze more thoughtful regulation. As long as ground-truthing remains clouded and abstract, society will struggle to prevent algorithms from causing harm and to optimize algorithms for the greater good.

Ground-truth datasets define AI algorithms’ fundamental goal of reliably predicting and generating a specific output—say, an image with requested specifications that resembles other input, such as web-crawled images. In other words, ground-truth datasets are deliberately constructed. As such, they, along with their resultant algorithms, are limited and arbitrary and bear the sociocultural fingerprints of the teams that made them…(More)”.

In shaping AI policy, stories about social impacts are just as important as expert information

Curated on Posted on by Stefaan Verhulst

Blog by Daniel S. Schiff and Kaylyn Jackson Schiff: “Will artificial intelligence (AI) save the world or destroy it? Will it lead to the end of manual labor and an era of leisure and luxury, or to more surveillance and job insecurity? Is it the start of a revolution in innovation that will transform the economy for the better? Or does it represent a novel threat to human rights?

Irrespective of what turns out to be the truth, what our key policymakers believe about these questions matters. It will shape how they think about the underlying problems that AI policy is aiming to address, and which solutions are appropriate to do so. …In late 2021, we ran a study to better understand the impact of policy narratives on the behavior of policymakers. We focused on US state legislators,…

In our analysis, we found something surprising. We measured whether legislators were more likely to engage with a message featuring a narrative or featuring expert information, which we assessed by seeing if they clicked on a given fact sheet/story or clicked to register for or attended the webinar.

Despite the importance attached to technical expertise in AI circles, we found that narratives were at least as persuasive as expert information. Receiving a narrative emphasizing, say, growing competition between the US and China, or the faulty arrest of Robert Williams due to facial recognition, led to a 30 percent increase in legislator engagement compared to legislators who only received basic information about the civil society organization. These narratives were just as effective as more neutral, fact-based information about AI with accompanying fact sheets…(More)”

Navigating the Metrics Maze: Lessons from Diverse Domains for Federal Chief Data Officers

Curated on Posted on by Stefaan Verhulst

Paper by the CDO Council: “In the rapidly evolving landscape of government, Federal Chief Data Officers (CDOs) have emerged as crucial leaders tasked with harnessing the power of data to drive organizational success. However, the relative newness of this role brings forth unique challenges, particularly in the realm of measuring and communicating the value of their efforts.

To address this measurement conundrum, this paper delves into lessons from non-data domains such as asset management, inventory management, manufacturing, and customer experience. While these fields share common ground with CDOs in facing critical questions, they stand apart in possessing established performance metrics. Drawing parallels with domains that have successfully navigated similar challenges offers a roadmap for establishing metrics that can transcend organizational boundaries.

By learning from the experiences of other domains and adopting a nuanced approach to metrics, CDOs can pave the way for a clearer understanding of the impact and value of their vital contributions to the data-driven future…(More)”.

How Tracking and Technology in Cars Is Being Weaponized by Abusive Partners

Curated on Posted on by Stefaan Verhulst

Article by Kashmir Hill: “After almost 10 years of marriage, Christine Dowdall wanted out. Her husband was no longer the charming man she had fallen in love with. He had become narcissistic, abusive and unfaithful, she said. After one of their fights turned violent in September 2022, Ms. Dowdall, a real estate agent, fled their home in Covington, La., driving her Mercedes-Benz C300 sedan to her daughter’s house near Shreveport, five hours away. She filed a domestic abuse report with the police two days later.

Her husband, a Drug Enforcement Administration agent, didn’t want to let her go. He called her repeatedly, she said, first pleading with her to return, and then threatening her. She stopped responding to him, she said, even though he texted and called her hundreds of times.

Ms. Dowdall, 59, started occasionally seeing a strange new message on the display in her Mercedes, about a location-based service called “mbrace.” The second time it happened, she took a photograph and searched for the name online.

“I realized, oh my God, that’s him tracking me,” Ms. Dowdall said.

“Mbrace” was part of “Mercedes me” — a suite of connected services for the car, accessible via a smartphone app. Ms. Dowdall had only ever used the Mercedes Me app to make auto loan payments. She hadn’t realized that the service could also be used to track the car’s location. One night, when she visited a male friend’s home, her husband sent the man a message with a thumbs-up emoji. A nearby camera captured his car driving in the area, according to the detective who worked on her case.

Ms. Dowdall called Mercedes customer service repeatedly to try to remove her husband’s digital access to the car, but the loan and title were in his name, a decision the couple had made because he had a better credit score than hers. Even though she was making the payments, had a restraining order against her husband and had been granted sole use of the car during divorce proceedings, Mercedes representatives told her that her husband was the customer so he would be able to keep his access. There was no button she could press to take away the app’s connection to the vehicle.

“This is not the first time that I’ve heard something like this,” one of the representatives told Ms. Dowdall…(More)”.

The 2010 Census Confidentiality Protections Failed, Here’s How and Why

Curated on Posted on by Stefaan Verhulst

Paper by John M. Abowd, et al: “Using only 34 published tables, we reconstruct five variables (census block, sex, age, race, and ethnicity) in the confidential 2010 Census person records. Using the 38-bin age variable tabulated at the census block level, at most 20.1% of reconstructed records can differ from their confidential source on even a single value for these five variables. Using only published data, an attacker can verify that all records in 70% of all census blocks (97 million people) are perfectly reconstructed. The tabular publications in Summary File 1 thus have prohibited disclosure risk similar to the unreleased confidential microdata. Reidentification studies confirm that an attacker can, within blocks with perfect reconstruction accuracy, correctly infer the actual census response on race and ethnicity for 3.4 million vulnerable population uniques (persons with nonmodal characteristics) with 95% accuracy, the same precision as the confidential data achieve and far greater than statistical baselines. The flaw in the 2010 Census framework was the assumption that aggregation prevented accurate microdata reconstruction, justifying weaker disclosure limitation methods than were applied to 2010 Census public microdata. The framework used for 2020 Census publications defends against attacks that are based on reconstruction, as we also demonstrate here. Finally, we show that alternatives to the 2020 Census Disclosure Avoidance System with similar accuracy (enhanced swapping) also fail to protect confidentiality, and those that partially defend against reconstruction attacks (incomplete suppression implementations) destroy the primary statutory use case: data for redistricting all legislatures in the country in compliance with the 1965 Voting Rights Act…(More)”.

Foundational Research Gaps and Future Directions for Digital Twins

Curated on Posted on by Stefaan Verhulst

Report by the National Academy of Engineering; National Academies of Sciences, Engineering, and Medicine: “Across multiple domains of science, engineering, and medicine, excitement is growing about the potential of digital twins to transform scientific research, industrial practices, and many aspects of daily life. A digital twin couples computational models with a physical counterpart to create a system that is dynamically updated through bidirectional data flows as conditions change. Going beyond traditional simulation and modeling, digital twins could enable improved medical decision-making at the individual patient level, predictions of future weather and climate conditions over longer timescales, and safer, more efficient engineering processes. However, many challenges remain before these applications can be realized.

This report identifies the foundational research and resources needed to support the development of digital twin technologies. The report presents critical future research priorities and an interdisciplinary research agenda for the field, including how federal agencies and researchers across domains can best collaborate…(More)”.

Conversing with Congress: An Experiment in AI-Enabled Communication

Curated on Posted on by Stefaan Verhulst

Blog by Beth Noveck: “Each Member of the US House Representative speaks for 747,184 people – a staggering increase from 50 years ago. In the Senate, this disproportion is even more pronounced: on average each Senator represents 1.6 million more constituents than her predecessor a generation ago. That’s a lower level of representation than any other industrialized democracy.  

As the population grows (over 60% since 1970), so, too, does constituent communications. 

But that communication is not working well. According to the Congressional Management Foundation, this overwhelming communication volume leads to dissatisfaction among voters who feel their views are not adequately considered by their representatives….A pioneering and important new study published in Government Information Quarterly entitled “Can AI communication tools increase legislative responsiveness and trust in democratic institutions?” (Volume 40, Issue 3, June 2023, 101829) from two Cornell researchers is shedding new light on the practical potential for AI to create more meaningful constituent communication….Depending on the treatment group they either were or were not told when replies were AI-drafted.

Their findings are telling. Standard, generic responses fare poorly in gaining trust. In contrast, all AI-assisted responses, particularly those with human involvement, significantly boost trust. “Legislative correspondence generated by AI with human oversight may be received favorably.” 

Screenshot 2023 12 12 at 4.21.16 Pm

While the study found AI-assisted replies to be more trustworthy, it also explored how the quality of these replies impacts perception. When they conducted this study, ChatGPT was still in its infancy and more prone to linguistic hallucinations so they also tested in a second experiment how people perceived higher, more relevant and responsive replies against lower quality, irrelevant replies drafted with AI…(More)”.

Using Data for Good: Identifying Who Could Benefit from Simplified Tax Filing

Curated on Posted on by Stefaan Verhulst

Blog by New America: “For years, New America Chicago has been working with state agencies, national and local advocates and thought leaders, as well as community members on getting beneficial tax credits, like the Earned Income Tax Credit (EITC) and Child Tax Credit (CTC), into the hands of those who need them most. Illinois paved the way recently with its innovative simplified filing initiative which helps residents easily claim their state Earned Income Credit (EIC) by confirming their refund with a prepopulated return.

This past year we had discussions with Illinois policymakers and state agencies, like the Illinois Department of Revenue (IDoR) and the Illinois Department of Human Services (IDHS), to envision new ways for expanding the simplified filing initiative. It is currently designed to reach those who have filed a federal tax return and claimed their EITC, leaving out non-filer households who typically do not file taxes because they earn less than the federal income requirement or have other barriers.

In Illinois, over 600,000 households are enrolled in SNAP, and over 1 million households are enrolled in Medicaid. Every year thousands of families spend countless hours applying for these and other social safety net programs using IDHS’ Application for Benefits Eligibility (ABE). Unfortunately, many of these households are most in need of the federal EITC and the recently expanded state EIC but will never receive it. We posed the question, what if Illinois could save families time and money by using that already provided income and household information to streamline access to the state EIC for low-income families that don’t normally file taxes?

Our friends at Inclusive Economy Lab (IEL) conducted analysis using Census microdata to estimate the number of Illinois households who are enrolled in Medicaid and SNAP but do not file their federal or state tax forms…(More)”.

Informing Decisionmakers in Real Time

Curated on Posted on by Stefaan Verhulst

Article by Robert M. Groves: “In response, the National Science Foundation (NSF) proposed the creation of a complementary group to provide decisionmakers at all levels with the best available evidence from the social sciences to inform pandemic policymaking. In May 2020, with funding from NSF and additional support from the Alfred P. Sloan Foundation and the David and Lucile Packard Foundation, NASEM established the Societal Experts Action Network (SEAN) to connect “decisionmakers grappling with difficult issues to the evidence, trends, and expert guidance that can help them lead their communities and speed their recovery.” We chose to build a network because of the widespread recognition that no one small group of social scientists would have the expertise or the bandwidth to answer all the questions facing decisionmakers. What was needed was a structure that enabled an ongoing feedback loop between researchers and decisionmakers. This structure would foster the integration of evidence, research, and advice in real time, which broke with NASEM’s traditional form of aggregating expert guidance over lengthier periods.

In its first phase, SEAN’s executive committee set about building a network that could both gather and disseminate knowledge. To start, we brought in organizations of decisionmakers—including the National Association of Counties, the National League of Cities, the International City/County Management Association, and the National Conference of State Legislatures—to solicit their questions. Then we added capacity to the network by inviting social and behavioral organizations—like the National Bureau of Economic Research, the National Hazards Center at the University of Colorado Boulder, the Kaiser Family Foundation, the National Opinion Research Center at the University of Chicago, The Policy Lab at Brown University, and Testing for America—to join and respond to questions and disseminate guidance. In this way, SEAN connected teams of experts with evidence and answers to leaders and communities looking for advice…(More)”.