Digital Ethics

How the vast amount of data collected from refugees is gathered, stored and shared today

Artificial intelligence (AI)–synthesized text, audio, image, and video are being weaponized for the purposes of nonconsensual intimate imagery, fin...

Autonomous vehicles (AVs) are often claimed to offer many societal benefits. Perhaps, the most important is the potential to save countless lives through anticipated improvements in safety by replacing human drivers with AI drivers. AVs will also dramatically...

Recent vision-language (VL) models are powerful, but can they reliably distinguish "right" from "left"? We curate three new corpora to quantify model comprehension of such basic spatial relations. These tests isolate spatial reasoning more precisely than existing datasets like VQAv2, e.g., our What'sUp benchmark contains sets of photographs varying only the spatial relations of objects, keeping their identity fixed (see Figure 1: models must comprehend not only the usual case of a dog under a table, but also, the same dog on top of the same table). We evaluate 18 VL models, finding that all perform poorly, e.g., BLIP finetuned on VQAv2, which nears human parity on VQAv2, achieves 56% accuracy on our benchmarks vs. humans at 99%. We conclude by studying causes of this surprising behavior, finding: 1) that popular vision-language pretraining corpora like LAION-2B contain little reliable data for learning spatial relationships; and 2) that basic modeling interventions like up-weighting preposition-containing instances or fine-tuning on our corpora are not sufficient to address the challenges our benchmarks pose. We are hopeful that these corpora will facilitate further research, and we release our data and code at https://github.com/amitakamath/whatsup_vlms.

One reason AI is tricky to regulate is that it cuts across everything

Software giant Adobe has been caught selling AI-generated images of the Israel-Hamas war, as spotted by Australian news outlet Crikey.

AI won't kill us all — but that doesn't make it trustworthy. Instead of getting distracted by future existential risks, AI ethics researcher Sasha Luccioni thinks we need to focus on the technology's current negative impacts, like emitting carbon, infringing copyrights and spreading biased information. She offers practical solutions to regulate our AI-filled future — so it's inclusive and transparent.

President Biden issued an Executive Order on AI with the goal of promoting “safe, secure, and trustworthy development and use of artificial intelligence.”

Statement asks government to help stop AI tools ‘using copyright-protected works with impunity’

AI image generators like Stable Diffusion and DALL-E amplify bias in gender, race and beyond, despite efforts to detoxify the data fueling these results.

Companies are training their generative AI models on vast swathes of the Internet—and there’s no real way to stop them

Is empathy the bedrock of morality? Two new studies suggest there is confusion around its meaning – and its usefulness in creating a more caring society

Low-skill cybercriminals are using a new tool to create hundreds of fake social media accounts in just a few seconds.

How can we ensure AI is free of bias? Dr. Joy Buolamwini, one of TIME’s Top 100 Most Influential People in AI and founder of The Algorithmic Justice League…

I hadn’t ever had cause to think about how manipulated online content could impact my life. Then, one winter morning, someone knocked at my door …

Today, President Biden issued a landmark Executive Order to ensure that America leads the way in seizing the promise and managing the risks of artificial intelligence (AI). The Executive Order establishes new standards for AI safety and security, protects Americans’ privacy, advances equity and civil rights, stands up for consumers and workers, promotes innovation and…

Machines writing dreary text, to be read by other machines

A scandal unfolding in Argentina shows the dangers of implementing facial recognition—even with laws and limits in place.

The pioneering AI researcher and activist shares her personal journey in a new book, and explains her concerns about today’s AI systems.

The Department of Veterans' Affairs system is complex, but it is the application form that turns that complexity into something unmanageable.

A leading researcher, who will attend this week’s AI safety summit in London, warns of ‘real threat to the public conversation’

Unchecked overuse is draining and damaging aquifers nationwide, a data investigation by the New York Times revealed, threatening millions of people and America’s status as a food superpower.

While educators worry that students are using generative AI to cheat, a new report finds that students are turning to the tool more for personal problems.

The Biden Administration will reportedly unveil a broad executive order on artificial intelligence next week. It’s allegedly scheduled for Monday, October 30.

While artificial intelligence (AI) seemed until recently to be science fiction, countless corporations across the globe are now researching ways to implement this technology in everyday life. AI works by processing [https://www.sas.com/en_us/insights/analytics/what-is-artificial-intelligence.html] large quantities of data, interpreting patterns in that data,

This summer, Meta began taking requests to delete data from its AI training. Artists say this new system is broken and fake. Meta says there is no opt-out program.

We tested ChatGPT-4’s ability to do a UX Audit of 12 webpages, and compared it to the results of 6 human UX professionals. GPT-4 had a 20% accuracy rate, 80% error rate, and discovered just 14–26% of the actual UX issues.

Researchers from MIT, Cohere for AI and 11 other institutions launched the Data Provenance Platform today in order to "tackle the data transparency crisis in the AI space."