Who Is Responsible for AI Copyright Infringement?

Curated on Posted on by Stefaan Verhulst

Article by Michael P. Goodyear: “Twenty-one-year-old college student Shane hopes to write a song for his boyfriend. In the past, Shane would have had to wait for inspiration to strike, but now he can use generative artificial intelligence to get a head start. Shane decides to use Anthropic’s AI chat system, Claude, to write the lyrics. Claude dutifully complies and creates the words to a love song. Shane, happy with the result, adds notes, rhythm, tempo, and dynamics. He sings the song and his boyfriend loves it. Shane even decides to post a recording to YouTube, where it garners 100,000 views.

But Shane did not realize that this song’s lyrics are similar to those of “Love Story,” Taylor Swift’s hit 2008 song. Shane must now contend with copyright law, which protects original creative expression such as music. Copyright grants the rights owner the exclusive rights to reproduce, perform, and create derivatives of the copyrighted work, among other things. If others take such actions without permission, they can be liable for damages up to $150,000. So Shane could be on the hook for tens of thousands of dollars for copying Swift’s song.

Copyright law has surged into the news in the past few years as one of the most important legal challenges for generative AI tools like Claude—not for the output of these tools but for how they are trained. Over two dozen pending court cases grapple with the question of whether training generative AI systems on copyrighted works without compensating or getting permission from the creators is lawful or not. Answers to this question will shape a burgeoning AI industry that is predicted to be worth $1.3 trillion by 2032.

Yet there is another important question that few have asked: Who should be liable when a generative AI system creates a copyright-infringing output? Should the user be on the hook?…(More)”

Launching the Data-Powered Positive Deviance Course

Curated on Posted on by Stefaan Verhulst

Blog by Robin Nowok: “Data-Powered Positive Deviance (DPPD) is a new method that combines the principles of Positive Deviance with the power of digital data and advanced analytics. Positive Deviance is based on the observation that in every community or organization, some individuals achieve significantly better outcomes than their peers, despite having similar challenges and resources. These individuals or groups are referred to as positive deviants.

The DPPD method follows the same logic as the Positive Deviance approach but leverages existing, non-traditional data sources, either instead of or in conjunction with traditional data sources. This allows for the identification of positive deviants on larger geographic and temporal scales. Once identified, we can then uncover the behaviors that lead to their success, enabling others to adopt these practices.

In a world where top-down solutions often fall short, DPPD offers a fresh perspective. It focuses on finding what’s already working within communities, rather than imposing external solutions. This can lead to more sustainable, culturally appropriate, and effective interventions.

Our online course is designed to get you started on your DPPD journey. Through five modules, you’ll gain both theoretical knowledge and practical skills to apply DPPD in your own work…(More)”.

Assessing potential future artificial intelligence risks, benefits and policy imperatives

Curated on Posted on by Stefaan Verhulst

OECD Report: “The swift evolution of AI technologies calls for policymakers to consider and proactively manage AI-driven change. The OECD’s Expert Group on AI Futures was established to help meet this need and anticipate AI developments and their potential impacts. Informed by insights from the Expert Group, this report distils research and expert insights on prospective AI benefits, risks and policy imperatives. It identifies ten priority benefits, such as accelerated scientific progress, productivity gains and better sense-making and forecasting. It discusses ten priority risks, such as facilitation of increasingly sophisticated cyberattacks; manipulation, disinformation, fraud and resulting harms to democracy; concentration of power; incidents in critical systems and exacerbated inequality and poverty. Finally, it points to ten policy priorities, including establishing clearer liability rules, drawing AI “red lines”, investing in AI safety and ensuring adequate risk management procedures. The report reviews existing public policy and governance efforts and remaining gaps…(More)”.

Human-AI coevolution

Curated on Posted on by Stefaan Verhulst

Paper by Dino Pedreschi et al: “Human-AI coevolution, defined as a process in which humans and AI algorithms continuously influence each other, increasingly characterises our society, but is understudied in artificial intelligence and complexity science literature. Recommender systems and assistants play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human choices through online platforms. The interaction between users and AI results in a potentially endless feedback loop, wherein users’ choices generate data to train AI models, which, in turn, shape subsequent user preferences. This human-AI feedback loop has peculiar characteristics compared to traditional human-machine interaction and gives rise to complex and often “unintended” systemic outcomes. This paper introduces human-AI coevolution as the cornerstone for a new field of study at the intersection between AI and complexity science focused on the theoretical, empirical, and mathematical investigation of the human-AI feedback loop. In doing so, we: (i) outline the pros and cons of existing methodologies and highlight shortcomings and potential ways for capturing feedback loop mechanisms; (ii) propose a reflection at the intersection between complexity science, AI and society; (iii) provide real-world examples for different human-AI ecosystems; and (iv) illustrate challenges to the creation of such a field of study, conceptualising them at increasing levels of abstraction, i.e., scientific, legal and socio-political…(More)”.

What is ‘sovereign AI’ and why is the concept so appealing (and fraught)?

Curated on Posted on by Stefaan Verhulst

Article by John Letzing: “Denmark unveiled its own artificial intelligence supercomputer last month, funded by the proceeds of wildly popular Danish weight-loss drugs like Ozempic. It’s now one of several sovereign AI initiatives underway, which one CEO believes can “codify” a country’s culture, history, and collective intelligence – and become “the bedrock of modern economies.”

That particular CEO, Jensen Huang, happens to run a company selling the sort of chips needed to pursue sovereign AI – that is, to construct a domestic vintage of the technology, informed by troves of homegrown data and powered by the computing infrastructure necessary to turn that data into a strategic reserve of intellect…

It’s not surprising that countries are forging expansive plans to put their own stamp on AI. But big-ticket supercomputers and other costly resources aren’t feasible everywhere.

Training a large language model has gotten a lot more expensive lately; the funds required for the necessary hardware, energy, and staff may soon top $1 billion. Meanwhile, geopolitical friction over access to the advanced chips necessary for powerful AI systems could further warp the global playing field.

Even for countries with abundant resources and access, there are “sovereignty traps” to consider. Governments pushing ahead on sovereign AI could risk undermining global cooperation meant to ensure the technology is put to use in transparent and equitable ways. That might make it a lot less safe for everyone.

An example: a place using AI systems trained on a local set of values for its security may readily flag behaviour out of sync with those values as a threat…(More)”.

Code and Craft: How Generative Ai Tools Facilitate Job Crafting in Software Development

Curated on Posted on by Stefaan Verhulst

Paper by Leonie Rebecca Freise et al: “The rapid evolution of the software development industry challenges developers to manage their diverse tasks effectively. Traditional assistant tools in software development often fall short of supporting developers efficiently. This paper explores how generative artificial intelligence (GAI) tools, such as Github Copilot or ChatGPT, facilitate job crafting—a process where employees reshape their jobs to meet evolving demands. By integrating GAI tools into workflows, software developers can focus more on creative problem-solving, enhancing job satisfaction, and fostering a more innovative work environment. This study investigates how GAI tools influence task, cognitive, and relational job crafting behaviors among software developers, examining its implications for professional growth and adaptability within the industry. The paper provides insights into the transformative impacts of GAI tools on software development job crafting practices, emphasizing their role in enabling developers to redefine their job functions…(More)”.

How to evaluate statistical claims

Curated on Posted on by Stefaan Verhulst

Blog by Sean Trott: “…The goal of this post is to distill what I take to be the most important, immediately applicable, and generalizable insights from these classes. That means that readers should be able to apply those insights without a background in math or knowing how to, say, build a linear model in R. In that way, it’ll be similar to my previous post about “useful cognitive lenses to see through”, but with a greater focus on evaluating claims specifically.

Lesson #1: Consider the whole distribution, not just the central tendency.

If you spend much time reading news articles or social media posts, the odds are good you’ll encounter some descriptive statistics: numbers summarizing or describing a distribution (a set of numbers or values in a dataset). One of the most commonly used descriptive statistics is the arithmetic mean: the sum of every value in a distribution, divided by the number of values overall. The arithmetic mean is a measure of “central tendency”, which just means it’s a way to characterize the typical or expected value in that distribution.

The arithmetic mean is a really useful measure. But as many readers might already know, it’s not perfect. It’s strongly affected by outliers—values that are really different from the rest of the distribution—and things like the skew of a distribution (see the image below for examples of skewed distribution).

Three different distributions. Leftmost is a roughly “normal” distribution; middle is a “right-skewed” distribution; and rightmost is a “left-skewed” distribution.

In particular, the mean is pulled in the direction of outliers or distribution skew. That’s the logic behind the joke about the average salary of people at a bar jumping up as soon as a billionaire walks in. It’s also why other measures of central tendency, such as the median, are often presented alongside (or instead of) the mean—especially for distributions that happen to be very skewed, such as income or wealth.

It’s not that one of these measures is more “correct”. As Stephen Jay Gould wrote in his article The Median Is Not the Message, they’re just different perspectives on the same distribution:

A politician in power might say with pride, “The mean income of our citizens is $15,000 per year.” The leader of the opposition might retort, “But half our citizens make less than $10,000 per year.” Both are right, but neither cites a statistic with impassive objectivity. The first invokes a mean, the second a median. (Means are higher than medians in such cases because one millionaire may outweigh hundreds of poor people in setting a mean, but can balance only one mendicant in calculating a median.)..(More)”

AI Analysis of Body Camera Videos Offers a Data-Driven Approach to Police Reform

Curated on Posted on by Stefaan Verhulst

Article by Ingrid Wickelgren: But unless something tragic happens, body camera footage generally goes unseen. “We spend so much money collecting and storing this data, but it’s almost never used for anything,” says Benjamin Graham, a political scientist at the University of Southern California.

Graham is among a small number of scientists who are reimagining this footage as data rather than just evidence. Their work leverages advances in natural language processing, which relies on artificial intelligence, to automate the analysis of video transcripts of citizen-police interactions. The findings have enabled police departments to spot policing problems, find ways to fix them and determine whether the fixes improve behavior.

Only a small number of police agencies have opened their databases to researchers so far. But if this footage were analyzed routinely, it would be a “real game changer,” says Jennifer Eberhardt, a Stanford University psychologist, who pioneered this line of research. “We can see beat-by-beat, moment-by-moment how an interaction unfolds.”

In papers published over the past seven years, Eberhardt and her colleagues have examined body camera footage to reveal how police speak to white and Black people differently and what type of talk is likely to either gain a person’s trust or portend an undesirable outcome, such as handcuffing or arrest. The findings have refined and enhanced police training. In a study published in PNAS Nexus in September, the researchers showed that the new training changed officers’ behavior…(More)”.

Access, Signal, Action: Data Stewardship Lessons from Valencia’s Floods

Curated on Posted on by Stefaan Verhulst

Article by Marta Poblet, Stefaan Verhulst, and Anna Colom: “Valencia has a rich history in water management, a legacy shaped by both triumphs and tragedies. This connection to water is embedded in the city’s identity, yet modern floods test its resilience in new ways.

During the recent floods, Valencians experienced a troubling paradox. In today’s connected world, digital information flows through traditional and social media, weather apps, and government alert systems designed to warn us of danger and guide rapid responses. Despite this abundance of data, a tragedy unfolded last month in Valencia. This raises a crucial question: how can we ensure access to the right data, filter it for critical signals, and transform those signals into timely, effective action?

Data stewardship becomes essential in this process.

In particular, the devastating floods in Valencia underscore the importance of:

  • having access to data to strengthen the signal (first mile challenges)
  • separating signal from noise
  • translating signal into action (last mile challenges)…(More)”.

Quantitative Urban Economics

Curated on Posted on by Stefaan Verhulst

Paper by Stephen J. Redding: “This paper reviews recent quantitative urban models. These models are sufficiently rich to capture observed features of the data, such as many asymmetric locations and a rich geography of the transport network. Yet these models remain sufficiently tractable as to permit an analytical characterization of their theoretical properties. With only a small number of structural parameters (elasticities) to be estimated, they lend themselves to transparent identification. As they rationalize the observed spatial distribution of economic activity within cities, they can be used to undertake counterfactuals for the impact of empirically-realistic public-policy interventions on this observed distribution. Empirical applications include estimating the strength of agglomeration economies and evaluating the impact of transport infrastructure improvements (e.g., railroads, roads, Rapid Bus Transit Systems), zoning and land use regulations, place-based policies, and new technologies such as remote working…(More)”.