If you've been keeping an eye the latest talent acquisition trends, you have probably seen a lot of buzz about recruiters using AI to eliminate hiring biases in hiring. However, as was recently proven by Amazon's hiring AI, if you're not feeding your AI the right data, discrimination will still show up. Read on to learn about how, when given the right data, AI can help you manage bias in your hiring decisions.
Use Your Data The Right Way To Eliminate Bias
As AI gets smarter, many talent acquisition specialists see it as a magic bullet for getting rid of discrimination. The truth is, even when your application of AI is unbiased, the data you're feeding it may not be. AI works best when you're using plenty of quality data to teach it, but, in Amazon's case, their computer models were taught to choose top applicants by learning from patterns found in resumes that had been submitted to the company over ten years. Most of these resumes came from men because the tech industry is male-dominated. Though Amazon taught their AI not to penalize gender-specific terms (i.e. "women's rowing team"), the machines still found ways to sort data in a discriminatory way. Mostly by using "soft" type data in applications rather than tightly controlled more objective choices.
The Amazon case goes to show that even huge tech companies can make mistakes, but these are mistakes we can learn from. When using AI to make hiring decisions, start with the data. Before feeding it to the program, clean the data, or at least be very selective about which data you use. Think about main bias points, like gender, age, and race eliminate terms and words that have psychographic indicators. That includes the way you construct sentences, by the way, because language gives away a LOT. This will (and should) take work and thought on your side, but in the end, being careful to make sure your data is chosen and managed well will teach your AI well.
There is no doubt that AI can be used to make *relatively* unbiased decisions, and as the technology continues to develop, it's likely to get more effective. To prepare for this, the most important thing you can do is make your data a priority. Contact us to see how SwoopTalent can help!
Photo by Mateus Campos Felipe on Unsplash
No Comments Yet
Let us know what you think