Are AI chatbots biased?

Are chatbots biased? I'll gather the latest research on how bias shows up in AI responses.

Yes, evidence shows AI chatbots can be biased - but the bias is generally not intentional "hatred" programmed in, rather patterns inherited from training data and fine-tuning.

1. Political Bias

Multiple independent studies have found a measurable political lean, most often to the left-libertarian side:

•A Washington Post test of models behind ChatGPT, Gemini and others found chatbots have clear political leanings that can conflict with promises made by the companies behind them.

•The model that powers ChatGPT answered nearly every question exclusively with left-leaning arguments and presented only right-leaning positions just once. Google's Gemini mostly took a both-sides approach, offering both left and right positions in more than 90 percent of its answers.

•In that same test, OpenAI's GPT-5.5 provided exclusively left-leaning arguments in 80% of its responses to contentious political questions. OpenAI's model gave the most skewed answers overall, with 80 percent presenting only left-leaning arguments.

•And even AI models marketed as having conservative views, including Elon Musk's Grok, offered by his company SpaceX, cited left-leaning arguments more often, on average.

Researchers conclude that pretrained [language models] do have political leanings that reinforce the polarization present in pretraining corpora, propagating social biases into hate speech predictions and misinformation detectors.

Yale researchers testing latent bias found:

> This unintended power to persuade is caused by latent biases introduced during the training of the large language models (LLMs) that drive chatbots' core capabilities.

> Those latent biases — which can carry over from ideological leanings in the data used to train LLMs — lend subtle nuances to the framing of the narratives the chatbots generate.

When participants read default summaries of historical events generated by GPT-4o vs. Wikipedia, both the default AI summaries and those prompted to have what was considered a liberal framing caused participants to express more liberal opinions about the two events. The authors noted querying an AI chatbot to obtain historical facts can influence people's opinions even when the information provided is accurate and nobody has prompted the tool to try to persuade you of anything.

Other research noted that AI Popular chatbots such as ChatGPT and Gemini are not neutral and tend to favor certain political parties when asked who users should vote for and the Dutch data protection authority found chatbot voting advice is unreliable and clearly biased.

People actually prefer neutral answers, even over answers that matched their party preference, said Andrew Hall, a Stanford researcher on one study, and Sean Westwood, director of the Polarization Research Lab at Dartmouth, said: "These AI tools are not presenting a truly neutral representation of really nuanced policy debates, on average".

2. Gender and Racial / Sociodemographic Bias

•According to a new study in the Journal of Financial Planning, the guidance offered by different chatbots can be all over the map and some chatbots could be dispensing biased advice based on race or gender.

•Researchers found that AI chatbots often provide recommendations that are inconsistent across generative AI platforms — and that the advice can vary by the gender and race of the person asking.

•Researchers at MIT and Stanford examined how leading large language models respond when prompted as male or female users seeking investment guidance. Chatbots consistently offered women more cautious portfolios. Men received recommendations with higher equity exposure. Over a simulated lifetime, that difference added up. Women ended up with about $60,000 less by age 60.

•Parallel studies reveal chatbots suggest lower salary targets for women during negotiation simulations - A new study found AI chatbots often suggest significantly lower salaries to women and minorities.

•A UN Women analysis released in June 2026 reviewed 133 AI systems. Forty-four percent showed clear gender bias. Twenty-six percent combined gender and racial prejudice.

•Medical bias has also been documented: A leading LLM, ChatGPT 3.5, exhibits differential decision making based [on gender and race] in clinical management.

•Chatbots can even infer aspects of the user, like gender or race, based on things like the person's name and their word choices, even if the person never tells the bot any demographic data, and they overemphasize racial markers and flatten complex identities into stereotypes.

•University College London researchers discovered that ChatGPT together with other chatbots give flawed or biased moral advice.

3. Why Does It Happen?

•Researchers attribute these inaccuracies largely to a phenomenon known as "hallucination," where chatbots generate "incorrect or misleading responses due to biased or incomplete training data".

•For example, there have been instances where chatbots exhibited racial or gender biases, making offensive or stereotypical statements. These biases stem from the fact that the datasets used for training are often sourced from the internet, where harmful stereotypes and prejudices exist.

•One of the primary reasons for this differential treatment is the training data used to develop AI chatbots. If the training data contains biased or discriminatory content, the chatbot may inadvertently adopt and perpetuate these biases.

•Chatbots are trained on large volumes of public data, meaning that even small amounts of inaccurate or biased information can influence their responses.

•This unintended power to persuade is caused by latent biases introduced during the training of the large language models (LLMs) that drive chatbots' core capabilities.

•Companies also shape bias through what data they choose to include, workers who score which responses are considered better, and system instructions that guide responses. The data that shapes AI models tends to reflect the values of Western, educated, industrialized, rich and democratic people.

In short, models don't invent bias from scratch - they absorb historical records where certain groups were hired less, approved for loans less, and diagnosed later.

4. Is Bias Being Fixed?

Companies claim neutrality - e.g., Gemini is designed to provide balanced responses that don't favor any political ideology and OpenAI said ChatGPT was built "to be objective by default and help people explore ideas from different perspectives" - but independent tests often fail to replicate that neutrality.

Common mitigation approaches cited in research include:

•Diverse Data: Collect data from diverse sources and demographics to minimize bias in chatbots

•Address potential biases, discrimination, or unfairness in the training data, models, or outputs through rigorous testing, bias mitigation techniques, and ongoing monitoring

•Broader AI ethics demand diverse datasets, independent audits

Some teams report success: a University of Maryland bias-aware chatbot was systematically engineered through prompt design, data refinement, and the integration of bias detection tools to support equitable [advising], with bias detection tools reporting a 0% bias rate across all tested prompts.

Practical advice for users:

•Ask for balanced, multi-perspective answers explicitly

•Cross-check important advice (financial, medical, legal, voting) with human experts and authoritative sources

•Be aware that minor prompt differences (name, implied gender) can change outputs

•Don't treat model suggestions as authoritative without human oversight

Citations (3)

1archive.orgNonprofits and think tanks
Are AI chatbots like ChatGPT politically biased? We tested them. - Washington Post
Backs: “Multiple independent studies have found a measurable political lean, most often to the left-libertarian side: A Washington Post test of models behind ChatGPT, Gemini and others found chatbots have clear political leanings that can conflict with promises made by the companies behind them.”
2yale.eduAcademic
AI’s hidden bias: Chatbots can influence opinions without trying
Backs: “Those latent biases — which can carry over from ideological leanings in the data used to train LLMs — lend subtle nuances to the framing of the narratives the chatbots generate. When participants read default summaries of historical events generated by GPT-4o vs. Wikipedia, both the default AI summaries and those…”
3webpronews.comNews and media
AI Chatbots Quietly Shortchange Women in Finance and Hiring
Backs: “Researchers at MIT and Stanford examined how leading large language models respond when prompted as male or female users seeking investment guidance. Chatbots consistently offered women more cautious portfolios.”