For chemists, the AI revolution has yet to happen

Curated on Posted on by Stefaan Verhulst

Editorial Team at Nature: “Many people are expressing fears that artificial intelligence (AI) has gone too far — or risks doing so. Take Geoffrey Hinton, a prominent figure in AI, who recently resigned from his position at Google, citing the desire to speak out about the technology’s potential risks to society and human well-being.

But against those big-picture concerns, in many areas of science you will hear a different frustration being expressed more quietly: that AI has not yet gone far enough. One of those areas is chemistry, for which machine-learning tools promise a revolution in the way researchers seek and synthesize useful new substances. But a wholesale revolution has yet to happen — because of the lack of data available to feed hungry AI systems.

Any AI system is only as good as the data it is trained on. These systems rely on what are called neural networks, which their developers teach using training data sets that must be large, reliable and free of bias. If chemists want to harness the full potential of generative-AI tools, they need to help to establish such training data sets. More data are needed — both experimental and simulated — including historical data and otherwise obscure knowledge, such as that from unsuccessful experiments. And researchers must ensure that the resulting information is accessible. This task is still very much a work in progress…(More)”.

The latest in homomorphic encryption: A game-changer shaping up

Curated on Posted on by Stefaan Verhulst

Article by Katharina Koerner: “Privacy professionals are witnessing a revolution in privacy technology. The emergence and maturing of new privacy-enhancing technologies that allow for data use and collaboration without sharing plain text data or sending data to a central location are part of this revolution.

The United Nations, the Organisation for Economic Co-operation and Development, the U.S. White House, the European Union Agency for Cybersecurity, the UK Royal Society, and Singapore’s media and privacy authorities all released reports, guidelines and regulatory sandboxes around the use of PETs in quick succession. We are in an era where there are high hopes for data insights to be leveraged for the public good while maintaining privacy principles and enhanced security.

A prominent example of a PET is fully homomorphic encryption, often mentioned in the same breath as differential privacy, federated learning, secure multiparty computation, private set intersection, synthetic data, zero knowledge proofs or trusted execution environments.

As FHE advances and becomes standardized, it has the potential to revolutionize the way we handle, protect and utilize personal data. Staying informed about the latest advancements in this field can help privacy pros prepare for the changes ahead in this rapidly evolving digital landscape.

Homomorphic encryption: A game changer?

FHE is a groundbreaking cryptographic technique that enables third parties to process information without revealing the data itself by running computations on encrypted data.

This technology can have far-reaching implications for secure data analytics. Requests to a databank can be answered without accessing its plain text data, as the analysis is conducted on data that remains encrypted. This adds a third layer of security for data when in use, along with protecting data at rest and in transit…(More)”.

China’s new AI rules protect people — and the Communist Party’s power

Curated on Posted on by Stefaan Verhulst

Article by Johanna M. Costigan: “In April, in an effort to regulate rapidly advancing artificial intelligence technologies, China’s internet watchdog introduced draft rules on generative AI. They cover a wide range of issues — from how data is trained to how users interact with generative AI such as chatbots. 

Under the new regulations, companies are ultimately responsible for the “legality” of the data they use to train AI models. Additionally, generative AI providers must not share personal data without permission, and must guarantee the “veracity, accuracy, objectivity, and diversity” of their pre-training data. 

These strict requirements by the Cyberspace Administration of China (CAC) for AI service providers could benefit Chinese users, granting them greater protections from private companies than many of their global peers. Article 11 of the regulations, for instance, prohibits providers from “conducting profiling” on the basis of information gained from users. Any Instagram user who has received targeted ads after their smartphone tracked their activity would stand to benefit from this additional level of privacy.  

Another example is Article 10 — it requires providers to employ “appropriate measures to prevent users from excessive reliance on generated content,” which could help prevent addiction to new technologies and increase user safety in the long run. As companion chatbots such as Replika become more popular, companies should be responsible for managing software to ensure safe use. While some view social chatbots as a cure for loneliness, depression, and social anxiety, they also present real risks to users who become reliant on them…(More)”.

As the Quantity of Data Explodes, Quality Matters

Curated on Posted on by Stefaan Verhulst

Article by Katherine Barrett and Richard Greene: “With advances in technology, governments across the world are increasingly using data to help inform their decision making. This has been one of the most important byproducts of the use of open data, which is “a philosophy- and increasingly a set of policies – that promotes transparency, accountability and value creation by making government data available to all,” according to the Organisation for Economic Co-operation and Development (OECD).

But as data has become ever more important to governments, the quality of that data has become an increasingly serious issue. A number of nations, including the United States, are taking steps to deal with it. For example, according to a study from Deloitte, “The Dutch government is raising the bar to enable better data quality and governance across the public sector.” In the same report, a case study about Finland states that “data needs to be shared at the right time and in the right way. It is also important to improve the quality and usability of government data to achieve the right goals.” And the United Kingdom has developed its Government Data Quality Hub to help public sector organizations “better identify their data challenges and opportunities and effectively plan targeted improvements.”

Our personal experience is with U.S. states and local governments, and in that arena the road toward higher quality data is a long and difficult one, particularly as the sheer quantity of data has grown exponentially. As things stand, based on our ongoing research into performance audits, it is clear that issues with data are impediments to the smooth process of state and local governments…(More)”.

Digital Anthropology Meets Data Science

Curated on Posted on by Stefaan Verhulst

Article by Katie Hillier: “Analyzing online ecosystems in real time, teams of anthropologists and data scientists can begin to understand rapid social changes as they happen.

Ask not what data science can do for anthropology, but what anthropology can do for data science. —Anders Kristian Munk, Why the World Needs Anthropologists Symposium 2022

In the last decade, emerging technologies, such as AI, immersive realities, and new and more addictive social networks, have permeated almost every aspect of our lives. These innovations are influencing how we form identities and belief systems. Social media influences the rise of subcultures on TikTok, the communications of extremist communities on Telegram, and the rapid spread of conspiracy theories that bounce around various online echo chambers. 

People with shared values or experiences can connect and form online cultures at unprecedented scales and speeds. But these new cultures are evolving and shifting faster than our current ability to understand them. 

To keep up with the depth and speed of online transformations, digital anthropologists are teaming up with data scientists to develop interdisciplinary methods and tools to bring the deep cultural context of anthropology to scales available only through data science—producing a surge in innovative methodologies for more effectively decoding online cultures in real time…(More)”.

Five Enablers for a New Phase of Behavioral Science

Curated on Posted on by Stefaan Verhulst

Article by Michael Hallsworth: “Over recent weeks I’ve been sharing parts of a “manifesto” that tries to give a coherent vision for the future of applied behavioral science. Stepping back, if I had to identify a theme that comes through the various proposals, it would be the need for self-reflective practice.

Behavioral science has seen a tremendous amount of growth and interest over the last decade, largely focused on expanding its uses and methods. My sense is it’s ready for a new phase of maturity. That maturity involves behavioral scientists reflecting on the various ways that their actions are shaped by structural, institutional, environmental, economic, and historical factors.

I’m definitely not exempt from this need for self-reflection. There are times when I’ve focused on a cognitive bias when I should have been spending more time exploring the context and motivations for a decision instead. Sometimes I’ve homed in on a narrow slice of a problem that we can measure, even if that means dispensing with wider systemic effects and challenges. Once I spent a long time trying to apply the language of heuristics and biases to explain why people were failing to use the urgent care alternatives to hospital emergency departments, before realizing that their behavior was completely reasonable.     

The manifesto critiques things like this, but it doesn’t have all the answers. Because it tries to both cover a lot of ground and go into detail, many of the hard knots of implementation go unpicked. The truth is that writing reports and setting goals is the easy part. Turning those goals into practice is much tougher; as behavioral scientists know, there is often a gap between intention and action.

Right now, I and others don’t always realize the ambitions set out in the manifesto. Changing that is going to take time and effort, and it will involve the discomfort of disrupting familiar practices. Some have made public commitments in this direction; my organization is working on upgrading its practices in line with proposals around making predictions prior to implementation, strengthening RCTs to cope with complexity, and enabling people to use behavioral science, among others.

The truth is that writing reports and setting goals is the easy part. Turning those goals into practice is much tougher; as behavioral scientists know, there is often a gap between intention and action.

But changes by individual actors will not be enough. The big issue is that several of the proposals require coordination. For example, one of the key ideas is the need for more multisite studies that are well coordinated and have clear goals. Another prioritizes developing international professional networks to support projects in low- and middle-income countries…(More)”.

Let’s Randomize America! 

Curated on Posted on by Stefaan Verhulst

Article by Dalton Conley: “…As our society has become less random, it has become more unequal. Many people know that inequality has been rising steadily over time, but a less-remarked-on development is that there’s been a parallel geographic shift, with high- and low-income people moving into separate, ever more distinct communities…As a sociologist, I study inequality and what can be done about it. It is, to say the least, a difficult problem to solve…I’ve come to believe that lotteries could help to crack this nut and make our society fairer and more equal. We can’t randomly assign where people live, of course. And we can’t integrate neighborhoods by fiat, either. We learned that lesson in the nineteen-seventies, when counties tried busing schoolchildren across town. Those programs aimed to create more racially and economically integrated schools; they resulted in the withdrawal of affluent students from urban public-school systems, and set off a political backlash that can still be felt today…

As a political tool, lotteries have come and gone throughout history. Sortition—the selection of political officials by lot—was first practiced in Athens in the sixth century B.C.E., and later reappeared in Renaissance city-states such as Florence, Venice, and Lombardy, and in Switzerland and elsewhere. In recent years, citizens’ councils—randomly chosen groups of individuals who meet to hammer out a particular issue, such as climate policy—have been tried in Canada, France, Iceland, Ireland, and the U.K. Some political theorists, such as Hélène Landemore, Jane Mansbridge, and the Belgian writer David Van Reybrouck, have argued that randomly selected decision-makers who don’t have to campaign are less likely to be corrupt or self-interested than those who must run for office; people chosen at random are also unlikely to be typically privileged, power-hungry politicians. The wisdom of the crowd improves when the crowd is more diverse…(More)”.

Data portability and interoperability: A primer on two policy tools for regulation of digitized industries

Curated on Posted on by Stefaan Verhulst

Article by Sukhi Gulati-Gilbert and Robert Seamans: “…In this article we describe two other tools, data portability and interoperability, that may be particularly useful in technology-enabled sectors. Data portability allows users to move data from one company to another, helping to reduce switching costs and providing rival firms with access to valuable customer data. Interoperability allows two or more technical systems to exchange data interactively. Due to its interactive nature, interoperability can help prevent lock-in to a specific platform by allowing users to connect across platforms. Data portability and interoperability share some similarities; in addition to potential pro-competitive benefits, the tools promote values of openness, transparency, and consumer choice.

After providing an overview of these topics, we describe the tradeoffs involved with implementing data portability and interoperability. While these policy tools offer lots of promise, in practice there can be many challenges involved when determining how to fund and design an implementation that is secure and intuitive and accomplishes the intended result.  These challenges require that policymakers think carefully about the initial implementation of data portability and interoperability. Finally, to better show how data portability and interoperability can increase competition in an industry, we discuss how they could be applied in the banking and social media sectors. These are just two examples of how data portability and interoperability policy could be applied to many different industries facing increased digitization. Our definitions and examples should be helpful to those interested in understanding the tradeoffs involved in using these tools to promote competition and innovation in the U.S. economy…(More)” See also: Data to Go: The Value of Data Portability as a Means to Data Liquidity.

German lawmakers mull creating first citizen assembly

Curated on Posted on by Stefaan Verhulst

APNews: “German lawmakers considered Wednesday whether to create the country’s first “citizen assembly’” to advise parliament on the issue of food and nutrition.

Germany’s three governing parties back the idea of appointing consultative bodies made up of members of the public selected through a lottery system who would discuss specific topics and provide nonbinding feedback to legislators. But opposition parties have rejected the idea, warning that such citizen assemblies risk undermining the primacy of parliament in Germany’s political system.

Baerbel Bas, the speaker of the lower house, or Bundestag, said that she views such bodies as a “bridge between citizens and politicians that can provide a fresh perspective and create new confidence in established institutions.”

“Everyone should be able to have a say,” Bas told daily Passauer Neue Presse. “We want to better reflect the diversity in our society.”

Environmental activists from the group Last Generation have campaigned for the creation of a citizen assembly to address issues surrounding climate change. However, the group argues that proposals drawn up by such a body should at the very least result in bills that lawmakers would then vote on.

Similar efforts to create citizen assemblies have taken place in other European countries such as Spain, Finland, Austria, Britain and Ireland…(More)”.

Misunderstanding Misinformation

Curated on Posted on by Stefaan Verhulst

Article by Claire Wardle: “In the fall of 2017, Collins Dictionary named fake news word of the year. It was hard to argue with the decision. Journalists were using the phrase to raise awareness of false and misleading information online. Academics had started publishing copiously on the subject and even named conferences after it. And of course, US president Donald Trump regularly used the epithet from the podium to discredit nearly anything he disliked.

By spring of that year, I had already become exasperated by how this term was being used to attack the news media. Worse, it had never captured the problem: most content wasn’t actually fake, but genuine content used out of context—and only rarely did it look like news. I made a rallying cry to stop using fake news and instead use misinformation, disinformation, and malinformation under the umbrella term information disorder. These terms, especially the first two, have caught on, but they represent an overly simple, tidy framework I no longer find useful.

Both disinformation and misinformation describe false or misleading claims, but disinformation is distributed with the intent to cause harm, whereas misinformation is the mistaken sharing of the same content. Analyses of both generally focus on whether a post is accurate and whether it is intended to mislead. The result? We researchers become so obsessed with labeling the dots that we can’t see the larger pattern they show.

By focusing narrowly on problematic content, researchers are failing to understand the increasingly sizable number of people who create and share this content, and also overlooking the larger context of what information people actually need. Academics are not going to effectively strengthen the information ecosystem until we shift our perspective from classifying every post to understanding the social contexts of this information, how it fits into narratives and identities, and its short-term impacts and long-term harms…(More)”.