What are the key privacy concerns associated with machine learning?
Machine Learning
Perspectives from experts on the questions that matter for Machine Learning
This article was an early beta test. See all-new collaborative articles about Machine Learning to get expert insights and join the conversation.
Machine learning can be a powerful data-driven approach to artificial intelligence, given its ability to teach computers to learn from data without being explicitly programmed. In today’s world, machine learning is already being used to address real-world challenges, from fraud detection to self-driving car tech. However, given its proximity to large amounts of data, machine learning can also raise a number of privacy-related concerns. Here are some of the biggest privacy concerns around machine learning.?
1. Indirect access to sensitive information: Machine learning algorithms can be used to make inferences about sensitive information, such as medical conditions or political views, from data that does not explicitly contain that information. For example, a machine learning algorithm might be able to infer that a person has diabetes from their purchase history.
“Your voice reveals more about you than you realize. To the human ear, the voice can instantly give away your mood, for example—it’s easy to tell if you’re excited or upset. But machines can learn a lot more: inferring your age, gender, ethnicity, socio-economic status, health conditions, and beyond. [...] Researchers are experimenting with multiple ways to enhance privacy for your voice. None of the methods are perfect, but they are being explored as possible ways to boost privacy in the infrastructure processing your voice data.
— Jimmy Orucevic is a cyber and digital risk manager at KPMG Switzerland. He holds over 10 years of experience in data protection, cybersecurity and IT law.?
2. Unwanted targeted advertising: Advertisers and retailers can use machine learning algorithms to predict what products or services a person is interested in, and in turn target them with ads for those products or services. At times, this can be intrusive and unwanted, and can lead to a feeling of being under surveillance.
3. Biased algorithms: If the data that is used to train a machine learning algorithm is biased, then the algorithm will be biased as well. For example, if an algorithm is trained on data that is predominantly male, it may end up exhibiting biases against women. This can have serious consequences, as machine learning is increasingly being used in decision-making, such as in hiring decisions or mortgage applications. Oftentimes, such biases can lead to the profiling and classification of people based on race or socioeconomic status.?
领英推荐
“Trying to insert sensitivity into people’s writing using (hyped) machine learning algorithms — that still have a long way to go before it can be used on its own without human intervention — is misguided. Also, gender bias in AI is not something that we can simply patch up with band-aid-like features. We need to challenge and rethink how we code AI.”
— Nermeen Ghoniem is an AI engineer at audio technology company Jabra. She holds a masters in human-centered artificial intelligence and holds over 4 years of experience in the technology industry.
4. Tracking applications: Machine learning can also be used to track people and their behavior. For example, a mobile app that uses machine learning could track a person’s location, the places they visit and the people they meet. This information could be used for a variety of purposes, such as targeted advertising or protection against identity theft. But users may also feel uncomfortable about information on their real-time whereabouts being revealed to third parties.
As artificial intelligence becomes a larger part of our everyday lives, the number of concerns related to privacy are also likely to increase. Moving forward, businesses should aim to offer transparency when it comes to when and how they are using machine learning, and users should be cautious if they do not wish for their personal data to be divulged.?
Explore more
How this article was made: An AI generated an initial answer to the question addressed in this article. The response was then fact checked, corrected, and amended by editor Felicia Hou . Any errors or additions? Please let us know in the comments.
Healthcare & Business Operations| Program/Product Leader & Strategist| Innovation, Chronic Disease Mitigation, Digital Health, & Value Based Care Advisor| Patient Advocate
2 年Machine learning can increase accuracy based on algorithms, likely reduce human errors, however, ML can also be used to steer people to better processes/decisions by building optimized algorithms based on research, ideal state, and a shift from known factors that affect results.?