{"id":99067,"date":"2021-03-30T06:26:27","date_gmt":"2021-03-30T06:26:27","guid":{"rendered":"https:\/\/dissidentvoice.org\/?p=114829"},"modified":"2021-03-30T06:26:27","modified_gmt":"2021-03-30T06:26:27","slug":"addressing-potential-bias-in-ai","status":"publish","type":"post","link":"https:\/\/radiofree.asia\/2021\/03\/30\/addressing-potential-bias-in-ai\/","title":{"rendered":"Addressing Potential Bias in AI"},"content":{"rendered":"

\"\"<\/a>Image Source: <\/em>Pixabay<\/em> <\/a><\/p>\n

Brookings defines artificial intelligence (AI) as<\/a> \u201ca wide-ranging branch of computer science concerned with building smart machines capable of performing tasks that typically require human intelligence.\u201d\u00a0 Replicating human intelligence in machines has positively influenced data collection, manufacturing processes, solving efficiency issues, and other business processes.<\/p>\n

Even with its various benefits<\/a>, AI has been running into some challenges when it comes to bias. It\u2019s important to continue researching these biases and implementing anything capable of ridding AI of discrimination completely.<\/p>\n

Here are three situations when AI is regularly biased but shouldn\u2019t be, guidance to move forward in each, and a bit more on why it’s essential to innovate AI so we can reap its benefits responsibly.<\/p>\n

Three Cases of AI Bias<\/strong><\/p>\n

Artificial intelligence is an excellent idea in theory. There\u2019s a thin line between responsible and irresponsible when it comes to using machines to predict future behavior based on past interactions, sift through data, identify critical information, or make decisions without emotional distraction.<\/p>\n

But with its continued use, scientists, data analysts, and developers are noticing some apparent biases that have to be addressed for AI to be used effectively.<\/p>\n

Here are three hurdles AI has been running into regarding bias and tips on how to overcome these challenges to leverage the benefits of artificial intelligence.<\/p>\n

Recruiting and Hiring<\/strong><\/p>\n

One of the most highlighted bias challenges in artificial intelligence is when it\u2019s used in hiring and recruiting processes. Chatbots, r\u00e9sum\u00e9-screening tools, and online assessments, among other tools, are all used to automate various hiring and recruiting strategies. Bias in recruiting and hiring processes is hugely detrimental to forming a diverse workforce.<\/p>\n

Your application may never make it past an Applicant Tracking System (ATS)<\/a> if the system\u2019s data is biased. Gender, names, and race have excluded perfectly qualified candidates from being invited for interviews because the AI system was trained this way. Eliminating bias in AI used for recruiting and hiring would ensure that every candidate is getting a fair shot at a position based on their qualifications versus being eliminated despite them.<\/p>\n

Recruiting and hiring processes should be personalized. Ensure that you\u2019ve found a balance between human influence and AI use to ensure candidates are consistently chosen based on the proper company criteria.<\/p>\n

Creation and Development Process<\/strong><\/p>\n

The creation and development process is largely where bias starts in AI. If the people who create and develop artificial intelligence machines, tools, and software are biased, they\u2019ll consciously or subconsciously program the system with that same bias.<\/p>\n

Artificial intelligence is only as good as the data inputted and the quality and fullness of the data collected. Those involved in the creation and development process should be required to detach their personal experiences from anything created at work.<\/p>\n

The AI field should be diversified<\/a> first to help dismantle any bias in the creation and development process. When bringing people on board for your implementation of AI, structure your hiring process to eliminate any candidates that display any significant bias, discriminatory, or racist behaviors and thought-processes. Ensure anyone you hire is committed only to diversity, change, and wholly supporting individuals across various cultures, races, ethnicities, and backgrounds.<\/p>\n

Social Media Algorithms<\/strong><\/p>\n

Billions of people in the world use social media. If you\u2019re one of those people, you know how vital algorithms are to the content we\u2019re showed and how our content shows up on other people\u2019s timelines and pages. When algorithms are biased, it adversely affects the relevancy of content offered and how influential you can become on these platforms.<\/p>\n

For example, in 2019, Facebook allowed its advertisers to intentionally target adverts<\/a> according to gender, race, and religion. Women were shown jobs geared toward nurturing roles and excluded from seeing job ads for masculine roles like janitors, drivers, and construction work. After discovering this bias in their options for targeted ads, they eliminated any ability to target individuals based on race, gender, or age in their ads.<\/p>\n

All social media platforms should follow Facebook\u2019s lead and be intentional about eliminating any ability to target people based on things like age, gender, race, and ethnicity. If you\u2019re running an ad of any sort, ensure they\u2019re rooted in diversity.<\/p>\n

Why it’s Important to Innovate AI<\/strong><\/p>\n

AI can help identify and reduce the impact of human bias. The benefits of artificial intelligence include<\/a>:<\/p>\n