FAQ
Why should I trust this site?
The Fairmind project is nonpartisan and noncommercial. We are not affiliated with any political party or partisan advocacy organization.
Our mission is to be fair to all sides. Realizing anyone can say that, we back it up with technology for enforcing accuracy and fairness.
Who is behind this site?
Fairmind is founded, funded, and powered by volunteers. There is no profit motive or partisan agenda.
Who is this site for?
Fairmind is for people who want to be informed and fair-minded about contested public issues. It is also for educators and students focused on civic engagement and critical thinking.
Why does helping people be fair-minded matter?
Democracy depends on citizens who can engage thoughtfully with contested issues. Our goal is to make it easier for people to do that.
For more, see Why Fairmind
Why is there only a single prototype decision guide?
We are developing and testing the guide format and related technologies. We are doing this with a sequential series of prototypes. When we confirm the guide format and authoring process, we will create decision guides on a variety of issues.
Can this only be used for ballot propositions?
It can be used for any policy issue with two opposing sides. In the future, we expect to cover a wide range of issues beyond ballot propositions.
Isn’t the current guide format a lot for someone to read?
The main content of the current prototype guide runs about 1,600 words—roughly a seven- or eight-minute read for most people. That doesn’t include the appendix on editorial choices or the footnotes, which aren’t meant to be read straight through (other than by AIs).
We believe this level of content is the minimum necessary. Some people will want to read all the main content, plus selected extras. Other people—we suspect most—will prefer an AI-guided walkthrough that condenses and personalizes the material to their answers.
New ways of slicing, summarizing, or interacting with the guide content may emerge over time. But these will still require the original content as a foundation. It is this foundation layer—the decision guide format and content as a kind of platform—where we are focused.
Isn’t it overly idealistic to expect people to care about being fair-minded?
We don’t assume everyone will care about being fair-minded. But for the people who do, we want to make it easier. Although these people may be a minority, they can play an outsized role in democratic decisions because they are among the persuadable middle, the swing voters, who often decide election outcomes.
Does Fairmind help address polarization?
Yes. Nearly two-thirds of Americans believe creating more accurate, trustworthy, accessible sources of news and information would help bring the country together.
In addition, Fairmind’s specific approach—presenting each side in its strongest form—causes people to develop more nuanced views. They may still disagree, but they’re less likely to view the other side as ignorant or malicious. This kind of fair-minded disagreement is a productive step away from polarization.
What’s the difference between using ChatGPT with Fairmind versus without?
When you use the Fairmind “Custom GPT,” ChatGPT has two things it normally does not have:
- The verified content of our decision guide
- Instructions for conducting a fair-minded decision process
This gives the AI a more reliable foundation for helping you than would be possible with its general training. Especially when helping people navigate contested issues, it is important that the AI not follow the patterns of public discourse, which are often distorted by misleading claims or the loudest voices.
Does Fairmind see my conversation with ChatGPT?
No. Your conversation is with ChatGPT on chatgpt.com. We do not see it.
Does the Fairmind prototype about Proposition 50 work with other AI assistants?
Yes, we’ve tested it with Claude, Gemini, and Grok. However, these services do not have the equivalent of ChatGPT’s “Custom GPT” capability. Instead, they require that you copy a long prompt from us and paste it into their interface. We decided that was too much friction for most users and so are focusing on the “Custom GPT” approach for now.
We are confident there will be better ways to integrate with a variety of AI assistants in the future—for example, with web browser sidebars that can read special instructions we provide for how to walk the user through one of our decision guide webpages.
Why don’t you integrate with AI assistants’ APIs?
If we did that, we would be absorbing the cost of every AI conversation by every user. As a noncommercial project, we cannot afford to do that at scale—at least not at today’s cost/quality levels for API integrations.
Our use of a “Custom GPT” is appealing because the costs are spread across the users, most of whom already have a free or paid ChatGPT account (and if not, can create one for free). This imposes minimal costs to individual users and no cost to us, yet it provides a familiar experience that utilizes a leading-edge AI model.
Is Fairmind’s content for a particular country?
Our focus is the United States. But people from elsewhere can still benefit from seeing the concept.
Are you claiming to have the objective view of issues you cover?
There is no singular objective view. In creating a decision guide, subjective choices are inevitable when choosing whether or not to cover an issue, how to frame the issue, which factors are most important to consider, which arguments are strongest, how to articulate the arguments, and so on.
Instead of pursuing the unrealistic goal of singular objectivity, we try to represent each issue in a fair and accurate way, which we subject to a variety of checks and reviews.
Is it even possible for you or anyone to be truly “fair”?
People are naturally biased in various ways, but that tendency can be countered.
Of course, the definition of “fairness” is itself subjective. For our purposes, fairness is what reasonable opponents would agree is a fair representation of their disagreement. (And if you question whether that is possible, imagine those opponents locked in a room. They can’t come out until they agree on a fair representation of the issue at hand. They don’t need agree on which side is right, only how to describe the disagreement. That they can do.)
Is there a risk of being overly fair, of conveying false equivalence?
False equivalence (or “bothsidesism”) happens when each side’s narrative is passed through equally and uncritically, without important context about the relative credibility of each side. It typically applies to quotes in a news article.
Fairmind decision guides have a different model than the reporting of a news article. For each issue, we examine the public discourse among advocates, experts, and observers. Then we filter it down to the best arguments and the verified evidence—with footnotes and analysis about credibility, conflicts of interest, and other contextual nuances. This approach is the opposite of the practices that sometimes cause false balance.
Also, to be clear, we will not waste anyone’s time on “controversies” where the evidence and arguments overwhelmingly support one view, like whether the Earth is round. We focus on issues where reasonable people disagree after examining reliable evidence and defensible arguments.
How do you handle potential biases in academic and scientific sources?
Our goal is to use the most reliable source(s) for any claim that could reasonably be questioned. Often, this means relying on scientific and academic sources, while recognizing that researchers can be influenced by political views, funders, career incentives, and personal ambitions.
When citing academic research, we evaluate each source based on several factors:
The researchers’ affiliations, reputations, and funding sources
The type of research (for example, single study, systematic review, or meta-analysis)
Whether it was peer-reviewed and where it was published
The strength of its methodology and quality of evidence
Whether its findings have been replicated by other researchers
How well-cited it is and whether it is controversial
Where relevant, we provide context about potential conflicts of interest, competing interpretations, or scholarly disagreements.
In summary: Rather than simply “trusting the science” (or scientists or academics), we focus on the quality of specific findings.
Why are there only two sides to each issue?
We recognize that many controversial issues have more than two sides. But in our testing, we’ve found that for many people, it’s already a mental challenge to compare arguments from two sides on multiple factors. If we add a third side, so every side needs to reference multiple others on multiple factors, it will be too much.
So, our approach has been to present two sides and to be careful to choose and frame issues in ways that fit that constraint.