Artificial Intelligence (AI) can easily address social and health inequalities. But there are ways to adjust the algorithms. Researchers proposed possible solutions at the last annual meeting of the American Association for the Advancement of Science (AAAS) in Washington in early March.
Maintenance cost…
In 2019, American researchers published Science A great example of the risks of using AI in health and, at the same time, the solutions to this problem. “It’s a common situation with AI,” said Emma Pearson, a computer scientist at Cornell Tech, a graduate institution at Cornell University in New York. »
Photo from Cornell Tech website
Emma Pearson is a computer scientist at Cornell Tech
The problem in America is that whites are more likely to have good insurance and therefore have access to care. Adjusted for cost distortion, the proportion of privileged black patients increased from 18% to 47%.
Emma Pearson is a computer scientist at Cornell Tech
… and gout pain
Mme Pearson has published himself on AI disruptions in healthcare. issue in 2021 Natural Medicine, he showed that AI analysis of x-ray images of the knee well replicated pain reports of osteoarthritis patients in whites, but not in blacks. “The less educated and poorer patients also had an underestimation of the severity of their arthritis,” says Ms.me Pearson. This reduced their access to knee surgery. However, in this case, the fix for the AI algorithm is not obvious.
Data base
One of the big problems is that the databases used to train AI programs have distortions with respect to society. “Often, whites are overrepresented in clinical trials,” says Ms.me Pearson. Lance Waller of Emory University in Atlanta, who organized the AAAS session, says clinical trials often involve healthy people. “We want to avoid interfering with drug-related data, background noise that prevents it from having statistical power,” says Waller, who specializes in correcting distortions in health databases. “We see the same problem with the use of AI in human resources. If AI is used for the initial skimming of applicants for a position in a predominantly male organization, women may automatically be left behind. We’ve even seen situations where keywords like “task completion” are overused on male CVs. »
custody
Photographic pictures
Forensic methods must take into account the sensibilities of the population that may be targeted by the police.
Police databases are particularly problematic, said Megan Price, president of the Human Rights Data Analysis Group (HRDAG) in San Francisco. “Certain types of crime are favored by the police, and certain neighborhoods, too,” Mr.me Price in his presentation. Efforts should be made to include the feelings and experiences of people who are often overlooked or over-targeted by the police in forensic algorithms. This may introduce distortions into AI’s assessment of recidivism risk, for example. » Mme Price is not against AI. “We used it to find mass graves of people executed as part of the war against drug traffickers in Mexico. »
Social websites
A possible solution to these database distortions is the use of AI to discover databases that have been overlooked by researchers. “One of my colleagues at Cornell Tech, Tanseem Chaudhary, is using AI to detect markers of certain diseases in social media posts,” says Mr.me Pearson. This is a great example of using AI to counter AI weaknesses. » The People Aware Computing Lab at Mme Chowdhury’s publications include diagnosis using “digital biomarkers”, social media keywords, mental fatigue, schizophrenia, persecution complex, chronic pain, anxiety and psychosis.
AI in the Third World
Photo from Harvard University website
Milind Tambe is a computer scientist at Harvard University and director of Google’s AI for Social Good program
Another session at the AAAS conference was very enthusiastic about the use of AI in healthcare. “I understand concerns about systemic inequalities in rich countries, but in many developing countries, and even among marginalized populations in industrialized countries, AI is already showing great promise,” said speaker Milind Tambe, a computer scientist at Harvard University and director of AI at Google. For social good projects. She cites the example of a program aimed at providing health information to poor mothers in India. “AI helped predict which mothers are at risk of dropping out of the program. We visit them at home to provide them with a personalized follow-up. »
Do you have a science question? Just ask Mathieu Perrault.
Learn more
-
- 50%
- Proportion of pharmaceutical companies planning to invest in AI in 2023
Source: Pharmaceutical Technology
- 60%
- Percentage of Americans who are uncomfortable with the idea of their doctor using AI for their care
Source: Pew Research Center
-
- 3
- Number of AI algorithms approved by the FDA in radiology in 2017
Source: Goldman Sachs
- 66
- Number of AI algorithms approved by the FDA in radiology in 2022
Source: Goldman Sachs
