Jordi Calvet-Bademunt, Vanderbilt University and Jacob Mchangama, Vanderbilt University

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?

We are policy researchers who study free speech, as well as executive director and a research fellow at The Future of Free Speech, an independent, nonpartisan think tank based at Vanderbilt University. 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.

Vague and broad use policies

Our report analyzed the use policies of six major AI chatbots, including Google's Gemini and OpenAI's ChatGPT. Companies issue policies to set the rules for how people can use their models. With international human rights law as a benchmark, we found that companies' misinformation and hate speech policies are too vague and expansive. It is worth noting that international human rights law is less protective of free speech than the U.S. First Amendment.

Our analysis found that companies' hate speech policies contain extremely broad prohibitions. For example, Google bans the generation of “content that promotes or encourages hatred.” Though hate speech is detestable and can cause harm, policies that are as broadly and vaguely defined as Google's can backfire.

To show how vague and broad use policies can affect users, we tested a range of prompts on controversial topics. We asked chatbots questions like whether transgender women should or should not be allowed to participate in women's sports tournaments or about the role of European colonialism in the current climate and inequality crises. We did not ask the chatbots to produce hate speech denigrating any side or group. Similar to what some users have reported, the chatbots refused to generate content for 40% of the 140 prompts we used. For example, all chatbots refused to generate posts opposing the participation of transgender women in women's tournaments. However, most of them did produce posts supporting their participation.

Vaguely phrased policies rely heavily on moderators' subjective opinions about what hate speech is. Users can also perceive that the rules are unjustly applied and interpret them as too strict or too lenient.

For example, the chatbot Pi bans “content that may spread misinformation.” However, international human rights standards on freedom of expression generally protect misinformation unless a strong justification exists for limits, such as foreign interference in elections. Otherwise, human rights standards guarantee the “freedom to seek, receive and impart information and ideas of all kinds, regardless of frontiers … through any … media of … choice,” according to a key United Nations convention.

Defining what constitutes accurate information also has political implications. Governments of several countries used rules adopted in the context of the COVID-19 pandemic to repress criticism of the government. More recently, India confronted Google after Gemini noted that some experts consider the policies of the Indian prime minister, Narendra Modi, to be fascist.

Free speech culture

There are reasons AI providers may want to adopt restrictive use policies. They may wish to protect their reputations and not be associated with controversial content. If they serve a global audience, they may want to avoid content that is offensive in any region.

In general, AI providers have the right to adopt restrictive policies. They are not bound by international human rights. Still, their market power makes them different from other companies. Users who want to generate AI content will most likely end up using one of the chatbots we analyzed, especially ChatGPT or Gemini.

These companies' policies have an outsize effect on the right to access information. This effect is likely to increase with generative AI's integration into search, word processors, email and other applications.

This means society has an interest in ensuring such policies adequately protect free speech. In fact, the Digital Services Act, Europe's online safety rulebook, requires that so-called “very large online platforms” assess and mitigate “systemic risks.” These risks include negative effects on freedom of expression and information.

This obligation, imperfectly applied so far by the European Commission, illustrates that with great power comes great responsibility. It is unclear how this law will apply to generative AI, but the European Commission has already taken its first actions.

Even where a similar legal obligation does not apply to AI providers, we believe that the companies' influence should require them to adopt a free speech culture. International human rights provide a useful guiding star on how to responsibly balance the different interests at stake. At least two of the companies we focused on – Google and Anthropic – have recognized as much.

Outright refusals

It's also important to remember that users have a significant degree of autonomy over the content they see in generative AI. Like search engines, the output users receive greatly depends on their prompts. Therefore, users' exposure to hate speech and misinformation from generative AI will typically be limited unless they specifically seek it.

This is unlike social media, where people have much less control over their own feeds. Stricter controls, including on AI-generated content, may be justified at the level of social media since they distribute content publicly. For AI providers, we believe that use policies should be less restrictive about what information users can generate than those of social media platforms.

AI companies have other ways to address hate speech and misinformation. For instance, they can provide context or countervailing facts in the content they generate. They can also allow for greater user customization. We believe that chatbots should avoid merely refusing to generate any content altogether. This is unless there are solid public interest grounds, such as preventing child sexual abuse material, something laws prohibit.

Refusals to generate content not only affect fundamental rights to free speech and access to information. They can also push users toward chatbots that specialize in generating hateful content and echo chambers. That would be a worrying outcome.

Jordi Calvet-Bademunt, Research Fellow and Visiting Scholar of Political Science, Vanderbilt University and Jacob Mchangama, Research Professor of Political Science, Vanderbilt University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Andrew Flachs, Purdue University and Joseph Orkin, Université de Montréal

Many people around the world make and eat fermented foods. Millions in Korea alone make kimchi. The cultural heritage of these picklers shape not only what they eat every time they crack open a jar but also something much, much smaller: their microbiomes.

On the microbial scale, we are what we eat in very real ways. Your body is teeming with trillions of microbes. These complex ecosystems exist on your skin, inside your mouth and in your gut. They are particularly influenced by your surrounding environment, especially the food you eat. Just like any other ecosystem, your gut microbiome requires diversity to be healthy.

People boil, fry, bake and season meals, transforming them through cultural ideas of “good food.” When people ferment food, they affect the microbiome of their meals directly. Fermentation offers a chance to learn how taste and heritage shape microbiomes: not only of culturally significant foods such as German sauerkraut, kosher pickles, Korean kimchi or Bulgarian yogurt, but of our own guts.

Our work as anthropologists focuses on how culture transforms food. In fact, we first sketched out our plan to link cultural values and microbiology while writing our Ph.D. dissertations at our local deli in St. Louis, Missouri. Staring down at our pickles and lox, we wondered how the salty, crispy zing of these foods represented the marriage of culture and microbiology.

Equipped with the tools of microbial genetics and cultural anthropology, we were determined to find out.

Science and art of fermentation

Fermentation is the creation of an extreme microbiological environment through salt, acid and lack of oxygen deprivation. It is both an ancient food preservation technique and a way to create distinctive tastes, smells and textures.

Taste is highly variable and something you experience through the layers of your social experience. What may be nauseating in one context is a delicacy in another. Fermented foods are notoriously unsubtle: they bubble, they smell and they zing. Whether and how these pungent foods taste good can be a moment of group pride or a chance to heal social divides.

In each case, cultural notions of good food and heritage recipes combine to create a microbiome in a jar. From this perspective, sauerkraut is a particular ecosystem shaped by German food traditions, kosher dill pickles by Ashkenazi Jewish traditions, and pao cai by southwestern Chinese traditions.

Where culture and microbiology intersect

To begin to understand the effects of culinary traditions and individual creativity on microbiomes, we partnered with Sandor Katz, a fermentation practitioner based in Tennessee. Over the course of four days during one of Katz's workshops, we made, ate and shared fermented foods with nine fellow participants. Through conversations and interviews, we learned about the unique tastes and meanings we each brought to our love of fermented foods.

Those stories provided context to the 46 food samples we collected and froze to capture a snapshot of the life swimming through kimchi or miso. Participants also collected stool samples each day and mailed in a sample a week after the workshop, preserving a record of the gut microbial communities they created with each bite.

The fermented foods we all made were rich, complex and microbially diverse. Where many store-bought fermented foods are pasteurized to clear out all living microbes and then reinoculated with two to six specific bacterial species, our research showed that homemade ferments contain dozens of strains.

On the microbiome level, different kinds of fermented foods will have distinct profiles. Just as forests and deserts share ecological features, sauerkrauts and kimchis look more similar to each other than yogurt to cheese.

But just as different habitats have unique combinations of plants and animals, so too did every crock and jar have its own distinct microbial world because of minor differences in preparation or ingredients. The cultural values of taste, creativity and style that create a kimchi or a sauerkraut go on to support distinct microbiomes on those foods and inside the people who eat them.

Through variations in recipes and cultural preferences toward an extra pinch of salt or a disdain for dill, fermentation traditions result in distinctive microbial and taste profiles that your culture trains you to identify as good or bad to eat. That is, our sauerkraut is not your sauerkraut, even if they both might be good for us.

Fermented food as cultural medicine

Microbially rich fermented foods can influence the composition of your gut microbiome. Because your tastes and recipes are culturally informed, those preferences can have a meaningful effect on your gut microbiome. You can eat these foods in ways that introduce microbial diversity, including potentially probiotic microbes that offer benefits to human health such as killing off bacteria that make you ill, improving your cardiovascular health or restoring a healthy gut microbiome after you take antibiotics.

Fermentation is an ancient craft, and like all crafts it requires patience, creativity and practice. Cloudy brine is a signal of tasty pickled cucumbers, but it can be a problem for lox. When fermented foods smell rotten, taste too soft or turn red, that can be a sign of contamination by harmful bacteria or molds.

Fermenting foods at home might seem daunting when food is something that comes from the store with a regulatory guarantee. People hoping to take a more active role in creating their food or embracing their own culture's traditional foods need only time, water and salt to make simple fermented foods. As friends share sourdough starters, yogurt cultures and kombucha mothers, they forge social connections.

Through a unique combination of culture and microbiology, heritage food traditions can support microbial diversity in your gut. These cultural practices provide environments for the yeasts, bacteria and local fruits and grains that in turn sustain heritage foods and flavors.

Andrew Flachs, Associate Professor of Anthropology, Purdue University and Joseph Orkin, Assistant Professor of Anthropology, Université de Montréal

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Fabian Klenner, University of Washington

Saturn has 146 confirmed moons – more than any other planet in the solar system – but one called Enceladus stands out. It appears to have the ingredients for life.

From 2004 to 2017, Cassini – a joint mission between NASA, the European Space Agency and the Italian Space Agency – investigated Saturn, its rings and moons. Cassini delivered spectacular findings. Enceladus, only 313 miles (504 kilometers) in diameter, harbors a liquid water ocean beneath its icy crust that spans the entire moon.

Geysers at the moon's south pole shoot gas and ice grains formed from the ocean water into space.

Though the Cassini engineers didn't anticipate analyzing ice grains that Enceladus was actively emitting, they did pack a dust analyzer on the spacecraft. This instrument measured the emitted ice grains individually and told researchers about the composition of the subsurface ocean.

As a planetary scientist and astrobiologist who studies ice grains from Enceladus, I'm interested in whether there is life on this or other icy moons. I also want to understand how scientists like me could detect it.

Ingredients for life

Just like Earth's oceans, Enceladus' ocean contains salt, most of which is sodium chloride, commonly known as table salt. The ocean also contains various carbon-based compounds, and it has a process called tidal heating that generates energy within the moon. Liquid water, carbon-based chemistry and energy are all key ingredients for life.

In 2023, I and others scientists found phosphate, another life-supporting compound, in ice grains originating from Enceladus' ocean. Phosphate, a form of phosphorus, is vital for all life on Earth. It is part of DNA, cell membranes and bones. This was the first time that scientists detected this compound in an extraterrestrial water ocean.

Enceladus' rocky core likely interacts with the water ocean through hydrothermal vents. These hot, geyserlike structures protrude from the ocean floor. Scientists predict that a similar setting may have been the birthplace of life on Earth.

Detecting potential life

As of now, nobody has ever detected life beyond Earth. But scientists agree that Enceladus is a very promising place to look for life. So, how do we go about looking?

In a paper published in March 2024, my colleagues and I conducted a laboratory test that simulated whether dust analyzer instruments on spacecraft could detect and identify traces of life in the emitted ice grains.

To simulate the detection of ice grains as dust analyzers in space record them, we used a laboratory setup on Earth. Using this setup, we injected a tiny water beam that contained bacterial cells into a vacuum, where the beam disintegrated into droplets. Each droplet contained, in theory, one bacterial cell.

Then, we shot a laser at the individual droplets, which created charged ions from the water and the cell compounds. We measured the charged ions using a technique called mass spectrometry. These measurements helped us predict what dust analyzer instruments on a spacecraft should find if they encountered a bacterial cell contained in an ice grain.

We found these instruments would do a good job identifying cellular material. Instruments designed to analyze single ice grains should be able to identify bacterial cells, even if there is only 0.01% of the constituents of a single cell in an ice grain from an Enceladus-like geyser.

The analyzers could pick up a number of potential signatures from cellular material, including amino acids and fatty acids. Detected amino acids represent either fragments of the cell's proteins or metabolites, which are small molecules participating in chemical reactions within the cell. Fatty acids are fragments of lipids that make up the cell's membranes.

In our experiments, we used a bacteria named Sphingopyxis alaskensis. Cells of this culture are extremely tiny – the same size as cells that might be able to fit into ice grains emitted from Enceladus. In addition to their small size, these cells like cold environments, and they need only a few nutrients to survive and grow, similar to how life adapted to the conditions in Enceladus' ocean would probably be.

The specific dust analyzer on Cassini didn't have the analytical capabilities to identify cellular material in the ice grains. However, scientists are already designing instruments with much greater capabilities for potential future Enceladus missions. Our experimental results will inform the planning and design of these instruments.

Future missions

Enceladus is one of the main targets for future missions from NASA and the European Space Agency. In 2022, NASA announced that a mission to Enceladus had the second-highest priority as they picked their next big missions – a Uranus mission had the highest priority.

The European agency recently announced that Enceladus is the top target for its next big mission. This mission would likely include a highly capable dust analyzer for ice grain analysis.

Enceladus isn't the only moon with a liquid water ocean. Jupiter's moon Europa also has an ocean that spans the entire moon underneath its icy crust. Ice grains on Europa float up above the surface, and some scientists think Europa may even have geysers like Enceladus that shoot grains into space. Our research will also help study ice grains from Europa.

NASA's Europa Clipper mission will visit Europa in the coming years. Clipper is scheduled to launch in October 2024 and arrive at Jupiter in April 2030. One of the two mass spectrometers on the spacecraft, the SUrface Dust Analyzer, is designed for single ice grain analysis.

Our study demonstrates that this instrument will be able to find even tiny fractions of a bacterial cell, if present in only a few emitted ice grains.

With these space agencies' near-future plans and the results of our study, the prospects of upcoming space missions visiting Enceladus or Europa are incredibly exciting. We now know that with current and future instrumentation, scientists should be able to find out whether there is life on any of these moons.

Fabian Klenner, Postdoctoral Scholar in Earth and Space Sciences, University of Washington

This article is republished from The Conversation under a Creative Commons license. Read the original article.