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 Fireflys 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 AIs political leanings and efforts to fight bias are important. Still, the conversation on AI ignores another crucial issue: What is the AI industrys 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.
Our report analyzed the use policies of six major AI chatbots, including Googles Gemini and OpenAIs 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 Googles 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 womens 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 womens 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.
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 AIs 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, Europes 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.
Its 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.
Go here to read the rest:
AI chatbots refuse to produce 'controversial' output why that's a free speech problem - The Conversation
- Bill Ackman says he's 'learned a lot' from Elon Musk's X - Quartz - April 22nd, 2024 [April 22nd, 2024]
- Opinion | Columbia, Free Speech and the Coddling of the American Right - The New York Times - April 22nd, 2024 [April 22nd, 2024]
- TikTok raises free speech concerns on bill passed by US House that may ban app - Voice of America - VOA News - April 22nd, 2024 [April 22nd, 2024]
- The Right Must Avoid the Left's Free Speech Pitfalls Minding The Campus - Minding The Campus - April 22nd, 2024 [April 22nd, 2024]
- Fear and loathing on America's college campuses as free speech is disappearing | Will Bunch - The Philadelphia Inquirer - April 22nd, 2024 [April 22nd, 2024]
- Harrison Ford Called 'Free Palestine' Supporters 'Force of Nature' in Speech? - Snopes.com - April 22nd, 2024 [April 22nd, 2024]
- Elon Musk to fund new First Amendment campaign to combat 'relentless attacks on free speech' - Fox News - April 22nd, 2024 [April 22nd, 2024]
- TikTok raises free speech concerns on bill passed by US House that may ban app - New York Post - April 22nd, 2024 [April 22nd, 2024]
- Navigating The Murky Waters Of Antisemitism, Free Speech, And Academic Freedom - Forbes - April 22nd, 2024 [April 22nd, 2024]
- The choice between safety and free speech is a false one - Daily Trojan Online - April 22nd, 2024 [April 22nd, 2024]
- UC Virtual Conference Centers Free Speech and Civil Rights Amid Ongoing Tensions on College Campuses - Diverse: Issues in Higher Education - April 22nd, 2024 [April 22nd, 2024]
- Free speech freeze-up | D.H. Robinson - The Critic - April 22nd, 2024 [April 22nd, 2024]
- Will Columbias law-school dean learn the law of free speech? - JNS.org - April 22nd, 2024 [April 22nd, 2024]
- OSU, OK State Regents for Higher Education complete first required free speech training - Daily O'Collegian - April 22nd, 2024 [April 22nd, 2024]
- TikTok uses free speech card to save itself from US ban, will it be enough? - Hindustan Times - April 22nd, 2024 [April 22nd, 2024]
- Settlement Reached in Free Speech Case at Temecula Valley Unified - ACLU of Southern California - April 22nd, 2024 [April 22nd, 2024]
- USC canceled its valedictorian speech: What the university got wrong. - Slate - April 22nd, 2024 [April 22nd, 2024]
- A free speech fiasco united the far-right here's why they remain divided - POLITICO Europe - April 22nd, 2024 [April 22nd, 2024]
- US TikTok Ban Bill Would 'Trample' On Free Speech Rights Of 170M Americans, Says Social Media Giant - Benzinga - April 22nd, 2024 [April 22nd, 2024]
- X marks the spot where free speech comes at a cost - Sydney Morning Herald - April 22nd, 2024 [April 22nd, 2024]
- Newly reinstated Texas Tech professor continues to advocate for free speech - KLBK | KAMC | EverythingLubbock.com - April 22nd, 2024 [April 22nd, 2024]
- Coin Center says Senate-presented stablecoin bill poses risks to innovation and free speech - crypto.news - April 22nd, 2024 [April 22nd, 2024]
- TikTok creators worry about free speech and income streams if ban succeeds: 'My livelihood is at stake' - CNBC - March 29th, 2024 [March 29th, 2024]
- As Texas students clash over Israel-Hamas war, Gov. Greg Abbott orders colleges to revise free speech policies - The Texas Tribune - March 29th, 2024 [March 29th, 2024]
- Opinion | The Debate Over Free Speech, Disinformation and Censorship - The New York Times - March 29th, 2024 [March 29th, 2024]
- VCU one of the top campuses in the country for free speech, advocacy group says - Axios - March 29th, 2024 [March 29th, 2024]
- Free Speech Is Under Attack in the U.S., but It's on the Ropes Elsewhere - Reason - March 29th, 2024 [March 29th, 2024]
- Free speech hangs in the balance in 3 Supreme Court cases - The Hill - March 29th, 2024 [March 29th, 2024]
- Free Speech Unmuted: Free Speech, Government Persuasion, and Government Coercion - Reason - March 29th, 2024 [March 29th, 2024]
- In crowded week for free speech, justices hear 3 First Amendment cases - Reporters Committee for Freedom of the Press - March 29th, 2024 [March 29th, 2024]
- Abbott Issues Guidance To Texas Colleges And Universities About Free Speech And Anti-Semitism - EastTexasRadio.com - March 29th, 2024 [March 29th, 2024]
- Gov. Abbott orders Texas universities to revise free speech policies to combat antisemitism - The UTD Mercury - March 29th, 2024 [March 29th, 2024]
- The Liberty Justice Center Urges the U.S. Supreme Court to Uphold Protections for Free Speech in Donor Disclosure ... - Liberty Justice Center - March 29th, 2024 [March 29th, 2024]
- GOP pushes anti-free speech bills to fight antisemitism - UnHerd - March 29th, 2024 [March 29th, 2024]
- PEN Union Cries Foul in Contract Talks as Criticism of PEN America Intensifies - Publishers Weekly - March 29th, 2024 [March 29th, 2024]
- CAIR-Texas Condemns Gov. Abbott's Anti-Palestinian Executive Order as Attack on Free Speech (Video) - - Council on American-Islamic Relations - March 29th, 2024 [March 29th, 2024]
- Dissent: When It Comes To Free Speech, the Editorial Board Is All Talk. | Opinion - Harvard Crimson - March 29th, 2024 [March 29th, 2024]
- Gov. Abbott calls for universities to update free speech policies, discipline violators to address antisemitism on campuses - The Daily Texan - March 29th, 2024 [March 29th, 2024]
- The Times Ed Board picks a confusing fight against the Emerald City Ride, free speech - Seattle Bike Blog - March 29th, 2024 [March 29th, 2024]
- Trump's Free Speech Defense on Trial in Georgia Election Interference Case - Hoodline - March 29th, 2024 [March 29th, 2024]
- Convicting Julian Assange Would Mean the End of Free Speech - The American Conservative - March 29th, 2024 [March 29th, 2024]
- Column: Banning TikTok is a blow to free speech - Redmond Spokesman - March 29th, 2024 [March 29th, 2024]
- Free Speech Is Under Such Threat In Canada It Would Make Orwell Blush - Forbes - March 29th, 2024 [March 29th, 2024]
- More on Coercion, Social Media, and Freedom of Speech: Rejoinder to Philip Hamburger - Reason - March 29th, 2024 [March 29th, 2024]
- OfS free speech guidance: time will tell if it builds understanding - The PIE News - March 29th, 2024 [March 29th, 2024]
- Kevin Rennie: Jaw-dropping attack on free speech and assembly in a CT town. It hurts us all. - Hartford Courant - March 2nd, 2024 [March 2nd, 2024]
- Chemerinsky: Navigating Free Speech on Campus, A First Amendment Perspective - The Collegian online - March 2nd, 2024 [March 2nd, 2024]
- Bentley Hosts Forum on Free Speech on College Campuses with Legal Expert Harvey Silverglate - Bentley University - March 2nd, 2024 [March 2nd, 2024]
- POLL: 69% of Americans believe country on wrong track on free speech - Foundation for Individual Rights in Education - March 2nd, 2024 [March 2nd, 2024]
- In Defense of Free Speech and the Mission of the University - Public Discourse - March 2nd, 2024 [March 2nd, 2024]
- Free Speech and Common Carriage: Unpacking the Supreme Court's Examination of the Texas and Florida Social ... - Public Knowledge - March 2nd, 2024 [March 2nd, 2024]
- So to Speak: The Free Speech Podcast | Free speech news: NetChoice, Taylor Swift, October 7, and Satan - Foundation for Individual Rights in Education - March 2nd, 2024 [March 2nd, 2024]
- Supreme Court to Decide How the First Amendment Applies to Social Media - The New York Times - March 2nd, 2024 [March 2nd, 2024]
- U.S. Supreme Court to weigh in on Texas social media law - The Texas Tribune - March 2nd, 2024 [March 2nd, 2024]
- Bill aimed at protecting free speech rights advancing in SC House - News From The States - March 2nd, 2024 [March 2nd, 2024]
- Champion of Free Speech and Journalism Margaret Talev Leads Institute for Democracy, Journalism and Citizenship ... - Syracuse University News - March 2nd, 2024 [March 2nd, 2024]
- Kinsey student says IU administrator infringed on free speech rights at demonstration - Indiana Daily Student - March 2nd, 2024 [March 2nd, 2024]
- CBS News boss who signed off on firing Catherine Herridge to get free speech award - New York Post - March 2nd, 2024 [March 2nd, 2024]
- U.S. Supreme Court to hear Texas and Florida cases about free speech and social media platforms - Texas Standard - March 2nd, 2024 [March 2nd, 2024]
- Should Honking Your Horn Be Considered Free Speech? - The Autopian - March 2nd, 2024 [March 2nd, 2024]
- Event: Free speech implications of the ICJ South Africa v. Israel case - ARTICLE 19 - Article 19 - March 2nd, 2024 [March 2nd, 2024]
- Takeaways From the Supreme Court Arguments on Social Media Laws - The New York Times - March 2nd, 2024 [March 2nd, 2024]
- Free Speech Unmuted: Book Bansor Are They? - Reason - March 2nd, 2024 [March 2nd, 2024]
- Florida anti-free speech bill targets 'liberal media' but guess who's really mad at it? - KeysNews.com - March 2nd, 2024 [March 2nd, 2024]
- Supreme Court arguments over social media laws and free speech are defining social media itself - Quartz - March 2nd, 2024 [March 2nd, 2024]
- Canadian measure would remove free speech protection for quoting Bible, sacred texts - Washington Times - March 2nd, 2024 [March 2nd, 2024]
- Suffield scraps plan to restrict the use of the town green following pushback from free speech advocates - FOX61 Hartford - March 2nd, 2024 [March 2nd, 2024]
- FIRST PERSON: Free speech fails for Zionists at UC Berkeley - The Jewish News of Northern California - March 2nd, 2024 [March 2nd, 2024]
- Menard Center and pre-law club host discussion regarding AI and Free Speech - UWEC Spectator - March 2nd, 2024 [March 2nd, 2024]
- Supreme Court arguments over future of social media and free speech - WFXRtv.com - March 2nd, 2024 [March 2nd, 2024]
- Judge skeptical of lawsuit brought by Elon Musk's X over hate speech research - NPR - March 2nd, 2024 [March 2nd, 2024]
- Israeli philosopher Yoram Hazony lectures on free speech, antisemitism while students hold vigil - Observer Online - March 2nd, 2024 [March 2nd, 2024]
- Big Tech fights Texas and Florida at SCOTUS, and Brett Kavanaugh might be the one saving the internet as we know it. - Slate - March 2nd, 2024 [March 2nd, 2024]
- Biden Is Trying to Balance Gaza Protests and Free Speech Rights as Demonstrators Disrupt His Events - U.S. News & World Report - January 29th, 2024 [January 29th, 2024]
- British Universities Are Repressing Free Speech on Palestine - Jacobin magazine - January 29th, 2024 [January 29th, 2024]
- The Future of Academic Freedom - The New Yorker - January 29th, 2024 [January 29th, 2024]
- "College Is All About Curiosity. And That Requires Free Speech." - Reason - January 29th, 2024 [January 29th, 2024]
- Palestine and the crisis of free speech on college campuses - The Real News Network - January 29th, 2024 [January 29th, 2024]
- College Is All About Curiosity. And That Requires Free Speech. - The New York Times - January 29th, 2024 [January 29th, 2024]
- Letters: Money is not free speech and a corporation is not a person - Akron Beacon Journal - January 29th, 2024 [January 29th, 2024]