Debugging Tech Journalism

Curated on Posted on by Stefaan Verhulst

Essay by Timothy B. Lee: “A huge proportion of tech journalism is characterized by scandals, sensationalism, and shoddy research. Can we fix it?

In November, a few days after Sam Altman was fired — and then rehired — as CEO of OpenAI, Reuters reported on a letter that may have played a role in Altman’s ouster. Several staffers reportedly wrote to the board of directors warning about “a powerful artificial intelligence discovery that they said could threaten humanity.”

The discovery: an AI system called Q* that can solve grade-school math problems.

“Researchers consider math to be a frontier of generative AI development,” the Reuters journalists wrote. Large language models are “good at writing and language translation,” but “conquering the ability to do math — where there is only one right answer — implies AI would have greater reasoning capabilities resembling human intelligence.”

This was a bit of a head-scratcher. Computers have been able to perform arithmetic at superhuman levels for decades. The Q* project was reportedly focused on word problems, which have historically been harder than arithmetic for computers to solve. Still, it’s not obvious that solving them would unlock human-level intelligence.

The Reuters article left readers with a vague impression that Q could be a huge breakthrough in AI — one that might even “threaten humanity.” But it didn’t provide readers with the context to understand what Q actually was — or to evaluate whether feverish speculation about it was justified.

For example, the Reuters article didn’t mention research OpenAI published last May describing a technique for solving math problems by breaking them down into small steps. In a December article, I dug into this and other recent research to help to illuminate what OpenAI is likely working on: a framework that would enable AI systems to search through a large space of possible solutions to a problem…(More)”.

AI chatbots refuse to produce ‘controversial’ output − why that’s a free speech problem

Curated on Posted on by Stefaan Verhulst

Article by Jordi Calvet-Bademunt and Jacob Mchangama: “Google recently made headlines globally because its chatbot Gemini generated images of people of color instead of white people in historical settings that featured white people. Adobe Firefly’s image creation tool saw similar issues. This led some commentators to complain that AI had gone “woke.” Others suggested these issues resulted from faulty efforts to fight AI bias and better serve a global audience.

The discussions over AI’s political leanings and efforts to fight bias are important. Still, the conversation on AI ignores another crucial issue: What is the AI industry’s approach to free speech, and does it embrace international free speech standards?…In a recent report, we found that generative AI has important shortcomings regarding freedom of expression and access to information.

Generative AI is a type of AI that creates content, like text or images, based on the data it has been trained with. In particular, we found that the use policies of major chatbots do not meet United Nations standards. In practice, this means that AI chatbots often censor output when dealing with issues the companies deem controversial. Without a solid culture of free speech, the companies producing generative AI tools are likely to continue to face backlash in these increasingly polarized times…(More)”.

What is ‘lived experience’?

Curated on Posted on by Stefaan Verhulst

Article by Patrick J Casey: “Everywhere you turn, there is talk of lived experience. But there is little consensus about what the phrase ‘lived experience’ means, where it came from, and whether it has any value. Although long used by academics, it has become ubiquitous, leaping out of the ivory tower and showing up in activism, government, consulting, as well as popular culture. The Lived Experience Leaders Movement explains that those who have lived experiences have ‘[d]irect, first-hand experience, past or present, of a social issue(s) and/or injustice(s)’. A recent brief from the US Department of Health and Human Services suggests that those who have lived experience have ‘valuable and unique expertise’ that should be consulted in policy work, since engaging those with ‘knowledge based on [their] perspective, personal identities, and history’ can ‘help break down power dynamics’ and advance equity. A search of Twitter reveals a constant stream of use, from assertions like ‘Your research doesn’t override my lived experience,’ to ‘I’m pretty sure you’re not allowed to question someone’s lived experience.’

A recurring theme is a connection between lived experience and identity. A recent nominee for the US Secretary of Labor, Julie Su, is lauded as someone who will ‘bring her lived experience as a daughter of immigrants, a woman of color, and an Asian American to the role’. The Human Rights Campaign asserts that ‘[l]aws and legislation must reflect the lived experiences of LGBTQ people’. An editorial in Nature Mental Health notes that incorporation of ‘people with lived experience’ has ‘taken on the status of a movement’ in the field.

Carried a step further, the notion of lived experience is bound up with what is often called identity politics, as when one claims to be speaking from the standpoint of an identity group – ‘in my lived experience as a…’ or, simply, ‘speaking as a…’ Here, lived experience is often invoked to establish authority and prompt deference from others since, purportedly, only members of a shared identity know what it’s like to have certain kinds of experience or to be a member of that group. Outsiders sense that they shouldn’t criticise what is said because, grounded in lived experience, ‘people’s spoken truths are, in and of themselves, truths.’ Criticism of lived experience might be taken to invalidate or dehumanise others or make them feel unsafe.

So, what is lived experience? Where did it come from? And what does it have to do with identity politics?…(More)”.

On the Manipulation of Information by Governments

Curated on Posted on by Stefaan Verhulst

Paper by Ariel Karlinsky and Moses Shayo: “Governmental information manipulation has been hard to measure and study systematically. We hand-collect data from official and unofficial sources in 134 countries to estimate misreporting of Covid mortality during 2020-21. We find that between 45%–55% of governments misreported the number of deaths. The lion’s share of misreporting cannot be attributed to a country’s capacity to accurately diagnose and report deaths. Contrary to some theoretical expectations, there is little evidence of governments exaggerating the severity of the pandemic. Misreporting is higher where governments face few social and institutional constraints, in countries holding elections, and in countries with a communist legacy…(More)”

The economic research policymakers actually need

Curated on Posted on by Stefaan Verhulst

Blog by Jed Kolko: “…The structure of academia just isn’t set up to produce the kind of research many policymakers need. Instead, top academic journal editors and tenure committees reward research that pushes the boundaries of the discipline and makes new theoretical or empirical contributions. And most academic papers presume familiarity with the relevant academic literature, making it difficult for anyone outside of academia to make the best possible use of them.

The most useful research often came instead from regional Federal Reserve banks, non-partisan think-tanks, the corporate sector, and from academics who had the support, freedom, or job security to prioritize policy relevance. It generally fell into three categories:

  1. New measures of the economy
  2. Broad literature reviews
  3. Analyses that directly quantify or simulate policy decisions.

If you’re an economic researcher and you want to do work that is actually helpful for policymakers — and increases economists’ influence in government — aim for one of those three buckets.

The pandemic and its aftermath brought an urgent need for data at higher frequency, with greater geographic and sectoral detail, and about ways the economy suddenly changed. Some of the most useful research contributions during that period were new data and measures of the economy: they were valuable as ingredients rather than as recipes or finished meals. Here are some examples:

The Formalization of Social Precarities

Curated on Posted on by Stefaan Verhulst

Anthology edited by Murali Shanmugavelan and Aiha Nguyen: “…explores platformization from the point of view of precarious gig workers in the Majority World. In countries like Bangladesh, Brazil, and India — which reinforce social hierarchies via gender, race, and caste — precarious workers are often the most marginalized members of society. Labor platforms made familiar promises to workers in these countries: work would be democratized, and people would have the opportunity to be their own boss. Yet even as platforms have upended the legal relationship between worker and employer, they have leaned into social structures to keep workers precarious — and in fact formalized those social precarities through surveillance and data collection…(More)”.

A Brief History of Automations That Were Actually People

Curated on Posted on by Stefaan Verhulst

Article by Brian Contreras: “If you’ve ever asked a chatbot a question and received nonsensical gibberish in reply, you already know that “artificial intelligence” isn’t always very intelligent.

And sometimes it isn’t all that artificial either. That’s one of the lessons from Amazon’s recent decision to dial back its much-ballyhooed “Just Walk Out” shopping technology, a seemingly science-fiction-esque software that actually functioned, in no small part, thanks to behind-the-scenes human labor.

This phenomenon is nicknamed “fauxtomation” because it “hides the human work and also falsely inflates the value of the ‘automated’ solution,” says Irina Raicu, director of the Internet Ethics program at Santa Clara University’s Markkula Center for Applied Ethics.

Take Just Walk Out: It promises a seamless retail experience in which customers at Amazon Fresh groceries or third-party stores can grab items from the shelf, get billed automatically and leave without ever needing to check out. But Amazon at one point had more than 1,000 workers in India who trained the Just Walk Out AI model—and manually reviewed some of its sales—according to an article published last year on the Information, a technology business website.

An anonymous source who’d worked on the Just Walk Out technology told the outlet that as many as 700 human reviews were needed for every 1,000 customer transactions. Amazon has disputed the Information’s characterization of its process. A company representative told Scientific American that while Amazon “can’t disclose numbers,” Just Walk Out has “far fewer” workers annotating shopping data than has been reported. In an April 17 blog post, Dilip Kumar, vice president of Amazon Web Services applications, wrote that “this is no different than any other AI system that places a high value on accuracy, where human reviewers are common.”…(More)”

The End of the Policy Analyst? Testing the Capability of Artificial Intelligence to Generate Plausible, Persuasive, and Useful Policy Analysis

Curated on Posted on by Stefaan Verhulst

Article by Mehrdad Safaei and Justin Longo: “Policy advising in government centers on the analysis of public problems and the developing of recommendations for dealing with them. In carrying out this work, policy analysts consult a variety of sources and work to synthesize that body of evidence into useful decision support documents commonly called briefing notes. Advances in natural language processing (NLP) have led to the continuing development of tools that can undertake a similar task. Given a brief prompt, a large language model (LLM) can synthesize information in content databases. This article documents the findings from an experiment that tested whether contemporary NLP technology is capable of producing public policy relevant briefing notes that expert evaluators judge to be useful. The research involved two stages. First, briefing notes were created using three models: NLP generated; human generated; and NLP generated/human edited. Next, two panels of retired senior public servants (with only one panel informed of the use of NLP in the experiment) were asked to judge the briefing notes using a heuristic evaluation rubric. The findings indicate that contemporary NLP tools were not able to, on their own, generate useful policy briefings. However, the feedback from the expert evaluators indicates that automatically generated briefing notes might serve as a useful supplement to the work of human policy analysts. And the speed with which the capabilities of NLP tools are developing, supplemented with access to a larger corpus of previously prepared policy briefings and other policy-relevant material, suggests that the quality of automatically generated briefings may improve significantly in the coming years. The article concludes with reflections on what such improvements might mean for the future practice of policy analysis…(More)”.

United against algorithms: a primer on disability-led struggles against algorithmic injustice

Curated on Posted on by Stefaan Verhulst

Report by Georgia van Toorn: “Algorithmic decision-making (ADM) poses urgent concerns regarding the rights and entitlements of people with disability from all walks of life. As ADM systems become increasingly embedded in government decision-making processes, there is a heightened risk of harm, such as unjust denial of benefits or inadequate support, accentuated by the expanding reach of state surveillance.

ADM systems have far reaching impacts on disabled lives and life chances. Despite this, they are often designed without the input of people with lived experience of disability, for purposes that do not align with the goals of full rights, participation, and justice for disabled people.

This primer explores how people with disability are collectively responding to the threats posed by algorithmic, data-driven systems – specifically their public sector applications. It provides an introductory overview of the topic, exploring the approaches, obstacles, and actions taken by people with disability in their ‘algoactivist’ struggles…(More)”.

The CFPB wants to rein in data brokers

Curated on Posted on by Stefaan Verhulst

Article by Gaby Del Valle: “The Consumer Financial Protection Bureau wants to propose new regulations that would require data brokers to comply with the Fair Credit Reporting Act. In a speech at the White House earlier this month, CFPB Director Rohit Chopra said the agency is looking into policies to “ensure greater accountability” for companies that buy and sell consumer data, in keeping with an executive order President Joe Biden issued in late February.

Chopra said the agency is considering proposals that would define data brokers that sell certain types of data as “consumer reporting agencies,” thereby requiring those companies to comply with the Fair Credit Reporting Act (FCRA). The statute bans sharing certain kinds of data (e.g., your credit report) with entities unless they serve a specific purpose outlined in the law (e.g., if the report is used for employment purposes or to extend a line of credit to someone).

The CFBP views the buying and selling of consumer data as a national security issue, not just a matter of privacy. Chopra mentioned three massive data breaches — the 2015 Anthem leak, the 2017 Equifax hack, and the 2018 Marriott breach — as examples of foreign adversaries illicitly obtaining Americans’ personal data. “When Americans’ health information, financial information, and even their travel whereabouts can be assembled into detailed dossiers, it’s no surprise that this raises risks when it comes to safety and security,” Chopra said. But the focus on high-profile hacks obscures a more pervasive, totally legal phenomenon: data brokers’ ability to sell detailed personal information to anyone who’s willing to pay for it…(More)”.