External Researcher Access to Closed Foundation Models

Curated on Posted on by Stefaan Verhulst

Report by Esme Harrington and Dr. Mathias Vermeulen: “…addresses a pressing issue: independent researchers need better conditions for accessing and studying the AI models that big companies have developed. Foundation models — the core technology behind many AI applications — are controlled mainly by a few major players who decide who can study or use them.

What’s the problem with access?

  • Limited access: Companies like OpenAI, Google and others are the gatekeepers. They often restrict access to researchers whose work aligns with their priorities, which means independent, public-interest research can be left out in the cold.
  • High-end costs: Even when access is granted, it often comes with a hefty price tag that smaller or less-funded teams can’t afford.
  • Lack of transparency: These companies don’t always share how their models are updated or moderated, making it nearly impossible for researchers to replicate studies or fully understand the technology.
  • Legal risks: When researchers try to scrutinize these models, they sometimes face legal threats if their work uncovers flaws or vulnerabilities in the AI systems.

The research suggests that companies need to offer more affordable and transparent access to improve AI research. Additionally, governments should provide legal protections for researchers, especially when they are acting in the public interest by investigating potential risks…(More)”.

Machines of Loving Grace

Curated on Posted on by Stefaan Verhulst

Essay by Dario Amodei: “I think and talk a lot about the risks of powerful AI. The company I’m the CEO of, Anthropic, does a lot of research on how to reduce these risks. Because of this, people sometimes draw the conclusion that I’m a pessimist or “doomer” who thinks AI will be mostly bad or dangerous. I don’t think that at all. In fact, one of my main reasons for focusing on risks is that they’re the only thing standing between us and what I see as a fundamentally positive future. I think that most people are underestimating just how radical the upside of AI could be, just as I think most people are underestimating how bad the risks could be.

In this essay I try to sketch out what that upside might look like—what a world with powerful AI might look like if everything goes right. Of course no one can know the future with any certainty or precision, and the effects of powerful AI are likely to be even more unpredictable than past technological changes, so all of this is unavoidably going to consist of guesses. But I am aiming for at least educated and useful guesses, which capture the flavor of what will happen even if most details end up being wrong. I’m including lots of details mainly because I think a concrete vision does more to advance discussion than a highly hedged and abstract one…(More)”.

G7 Toolkit for Artificial Intelligence in the Public Sector

Curated on Posted on by Stefaan Verhulst

OECD Toolkit: “…a comprehensive guide designed to help policymakers and public sector leaders translate principles for safe, secure, and trustworthy Artificial Intelligence (AI) into actionable policies. AI can help improve the efficiency of internal operations, the effectiveness of policymaking, the responsiveness of public services, and overall transparency and accountability. Recognising both the opportunities and risks posed by AI, this toolkit provides practical insights, shares good practices for the use of AI in and by the public sector, integrates ethical considerations, and provides an overview of G7 trends. It further showcases public sector AI use cases, detailing their benefits, as well as the implementation challenges faced by G7 members, together with the emerging policy responses to guide and coordinate the development, deployment, and use of AI in the public sector. The toolkit finally highlights key stages and factors characterising the journey of public sector AI solutions…(More)”.

Deliberative Technology: Designing AI and Computational Democracy for Peacebuilding in Highly-Polarized Contexts

Curated on Posted on by Stefaan Verhulst

Report by Lisa Schirch: “This is a report on an international workshop for 45 peacebuilders, co-hosted by Toda Peace Institute and the University of Notre Dame’s Kroc Institute for International Peace Studies in June 2024.  Emphasizing citizen participation and collective intelligence, the workshop explored the intersection of digital democracy and algorithmic technologies designed to enhance democratic processes. Central to the discussions were deliberative technologies, a new class of tools that facilitate collective discussion and decision-making by incorporating both qualitative and quantitative inputs, supported by bridging algorithms and AI. The workshop provided a comprehensive overview of how these innovative approaches and technologies can contribute to more inclusive and effective democratic processes, particularly in contexts marked by polarization and conflict…(More)”

Data’s Role in Unlocking Scientific Potential

Curated on Posted on by Stefaan Verhulst

Report by the Special Competitive Studies Project: “…we outline two actionable steps the U.S. government can take immediately to address the data sharing challenges hindering scientific research.

1. Create Comprehensive Data Inventories Across Scientific Domains

We recommend the Secretary of Commerce, acting through the Department of Commerce’s Chief Data Officer and the Director of the National Institute of Standards and Technology (NIST), and with the Federal Chief Data Officer Council (CDO Council) create a government-led inventory where organizations – universities, industries, and research institutes – can catalog their datasets with key details like purpose, description, and accreditation. Similar to platforms like data.gov, this centralized repository would make high-quality data more visible and accessible, promoting scientific collaboration. To boost participation, the government could offer incentives, such as grants or citation credits for researchers whose data is used. Contributing organizations would also be responsible for regularly updating their entries, ensuring the data stays relevant and searchable. 

2. Create Scientific Data Sharing Public-Private Partnerships

A critical recommendation of the National Data Action Plan was for the United States to facilitate the creation of data sharing public-private partnerships for specific sectors. The U.S. Government should coordinate data sharing partnerships with its departments and agencies, industry, academia, and civil society. Data collected by one entity can be tremendously valuable to others. But incentivizing data sharing is challenging as privacy, security, legal (e.g., liability), and intellectual property (IP) concerns can limit willingness to share. However, narrowly-scoped PPPs can help overcome these barriers, allowing for greater data sharing and mutually beneficial data use…(More)”

Can LLMs advance democratic values?

Curated on Posted on by Stefaan Verhulst

Paper by Seth Lazar and Lorenzo Manuali: “LLMs are among the most advanced tools ever devised for analysing and generating linguistic content. Democratic deliberation and decision-making involve, at several distinct stages, the production and analysis of language. So it is natural to ask whether our best tools for manipulating language might prove instrumental to one of our most important linguistic tasks. Researchers and practitioners have recently asked whether LLMs can support democratic deliberation by leveraging abilities to summarise content, as well as to aggregate opinion over summarised content, and indeed to represent voters by predicting their preferences over unseen choices. In this paper, we assess whether using LLMs to perform these and related functions really advances the democratic values that inspire these experiments. We suggest that the record is decidedly mixed. In the presence of background inequality of power and resources, as well as deep moral and political disagreement, we should be careful not to use LLMs in ways that automate non-instrumentally valuable components of the democratic process, or else threaten to supplant fair and transparent decision-making procedures that are necessary to reconcile competing interests and values. However, while we argue that LLMs should be kept well clear of formal democratic decision-making processes, we think that they can be put to good use in strengthening the informal public sphere: the arena that mediates between democratic governments and the polities that they serve, in which political communities seek information, form civic publics, and hold their leaders to account…(More)”.

The Age of AI Nationalism and Its Effects

Curated on Posted on by Stefaan Verhulst

Paper by Susan Ariel Aaronson: “Policy makers in many countries are determined to develop artificial intelligence (AI) within their borders because they view AI as essential to both national security and economic growth. Some countries have proposed adopting AI sovereignty, where the nation develops AI for its people, by its people and within its borders. In this paper, the author makes a distinction between policies designed to advance domestic AI and policies that, with or without direct intent, hamper the production or trade of foreign-produced AI (known as “AI nationalism”). AI nationalist policies in one country can make it harder for firms in another country to develop AI. If officials can limit access to key components of the AI supply chain, such as data, capital, expertise or computing power, they may be able to limit the AI prowess of competitors in country Y and/or Z. Moreover, if policy makers can shape regulations in ways that benefit local AI competitors, they may also impede the competitiveness of other nations’ AI developers. AI nationalism may seem appropriate given the import of AI, but this paper aims to illuminate how AI nationalistic policies may backfire and could divide the world into AI haves and have nots…(More)”.

Social Systems Evidence

Curated on Posted on by Stefaan Verhulst

About: “…a continuously updated repository of syntheses of research evidence about the programs, services and products available in a broad range of government sectors and program areas (e.g., climate action, community and social services, economic development and growth, education, environmental conservation, education, housing and transportation) as well as the governance, financial and delivery arrangements within which these programs, services and products are provided, and the implementation strategies that can help to ensure that these programs, services and products get to those who need them. 

The content covers the Sustainable Development Goals, with the exceptions of the health part of goal 3 (which is already well covered by existing databases).

The types of syntheses include evidence briefs for policy, overviews of evidence syntheses, evidence syntheses addressing questions about effectiveness, evidence syntheses addressing other types of questions, evidence syntheses in progress (i.e., protocols for evidence syntheses), and evidence syntheses being planned (i.e., registered titles for evidence syntheses). Social Systems Evidence also contains a continuously updated repository of economic evaluations in these same domains…(More)”

First-of-its-kind dataset connects greenhouse gases and air quality

Curated on Posted on by Stefaan Verhulst

NOAA Research: “The GReenhouse gas And Air Pollutants Emissions System (GRA²PES), from NOAA and the National Institute of Standards and Technology (NIST), combines information on greenhouse gas and air quality pollutant sources into a single national database, offering innovative interactive map displays and new benefits for both climate and public health solutions.

A new U.S.-based system to combine air quality and greenhouse gas pollution sources into a single national research database is now available in the U.S. Greenhouse Gas Center portal. This geospatial data allows leaders at city, state, and regional scales to more easily identify and take steps to address air quality issues while reducing climate-related hazards for populations.

The dataset is the GReenhouse gas And Air Pollutants Emissions System (GRA²PES). A research project developed by NOAA and NIST, GRA²PES captures monthly greenhouse gas (GHG) emissions activity for multiple economic sectors to improve measurement and modeling for both GHG and air pollutants across the contiguous U.S.

Having the GHG and air quality constituents in the same dataset will be exceedingly helpful, said Columbia University atmospheric scientist Roisin Commane, the lead on a New York City project to improve emissions estimates…(More)”.

Harnessing the feed: social media for mental health information and support 

Curated on Posted on by Stefaan Verhulst

Report by ReachOut: “…highlights how a social media ban could cut young people off from vital mental health support, including finding that 73 per cent of young people in Australia turn to social media when it comes to support for their mental health.

Based on research with over 2000 young people, the report found a range of benefits for young people seeking mental health support via social media (predominantly TikTok, YouTube and Instagram). 66 per cent of young people surveyed reported increased awareness about their mental health because of relevant content they accessed via social media, 47 per said they had looked for information about how to get professional mental health support on social media and 40 per cent said they sought professional support after viewing mental health information on social media. 

Importantly, half of young people with a probable mental health condition said that they were searching for mental health information or support on social media because they don’t have access to professional support. 

However, young people also highlighted a range of concerns about social media via the research. 38 per cent were deeply concerned about harmful mental health content they have come across on platforms and 43 per cent of the young people who sought support online were deeply concerned about the addictive nature of social media.  

The report highlights young people’s calls for social media to be safer. They want: an end to addictive features like infinite scroll, more control over the content they see, better labelling of mental health information from credible sources, better education and more mental health information provided across platforms…(More)”.