An unbiased, purely fact-based AI chatbot is a cute idea, but it's technically impossible. To understand why, it's worth reading a story I just published on new research that sheds light on how political bias creeps into AI language systems. Researchers conducted tests on 14 large language models and found that OpenAI's ChatGPT and GPT-4 were the most left-wing libertarian, while Meta's LLaMA was the most right-wing authoritarian. "We believe no language model can be entirely free from political biases," Chan Park, a PhD researcher at Carnegie Mellon University, who was part of the study, told me. One of the most pervasive myths around AI is that the technology is neutral and unbiased.

A seagull attacking a man with spaghetti, a skateboarding dinosaur, and a cup of coffee that contains the universe. These are just some of the odd prompts that people have given to new AI systems, for an often unusual, yet incredibly detailed image in return. Whilst majority of the AI art you are likely seeing on social media comes from Open-A-I's, DALL E mini, other notable artists are Open-A-I's DALL-E 2, and Google Research's Imagen Google research shows the technology appears to involve "several social biases and stereotypes". But experts fear these systems are also capable of producing disinformation based off the gender and cultural biases from the data they feed off. An OpenAI online document titled'Risks and Limitations', which shows these biases with an example of how a text description of a CEO, for instance, only shows images of predominantly white men. Technology has "an overall bias towards generating images of people with lighter skin tones and a tendency for images portraying different professions to align with Western gender stereotypes."

Artificial intelligence and medical algorithms are deeply intertwined with our modern health care system. These technologies mimic the thought processes of doctors to make medical decisions and are designed to help providers determine who needs care. But one big problem with artificial intelligence is that it very often replicate the biases and blind spots of the humans who create them. Researchers and physicians have warned that algorithms used to determine who gets kidney transplants, heart surgeries and breast cancer diagnoses display racial bias. Those problems can lead to detrimental care that, in some cases, can jeopardize the health of millions of patients.

In this brief animated video, Adam Pah, a clinical assistant professor of management and organizations at the Kellogg School, explains why leaders need to be thoughtful about the data they use--and how they put that data to use--to avoid problems with bias when designing artificial intelligence and machine learning programs.

Yet the inexperienced or rushed data scientist skipped past feature engineering, the critical stage at which those invalid fields would have been removed. The experienced data scientist would know to invest lots of time in feature engineering to explicitly screen out potential bias from our training data. If our hiring data to date has a past human bias of not hiring women at the same rate as men, our machine learning model would learn to emulate that behavior unless we explicitly removed gender from consideration. It's easy to see how bias could creep in if inexperienced or rushed data scientists are building models from massive datasets.

The lesson of Trump's victory is not that data is dead. The lesson is that data is flawed. It has always been flawed--and always will be. Before Donald Trump won the presidency on Tuesday night, everyone from Nate Silver to The New York Times to CNN predicted a Trump loss--and by sizable margins. "The tools that we would normally use to help us assess what happened failed," Trump campaign reporter Maggie Haberman said in the Times. As Haberman explained, this happened on both sides of the political divide.

Computers are only as good, or as bad, as the people who program them. And it turns out that many individuals who create machine learning algorithms are presumably and unintentionally building in race and gender bias. In part one of a two-part interview, Richard Sharp, CTO of predictive marketing company Yieldify, explains how it happens. The Enterprisers Project (TEP): Machines are genderless, have no race, and are in and of themselves free of bias. How does bias creep in?