2022_심화영어독해와작문

121 Just like humans, AI systems are also expected to follow social norms, and to be fair and unbiased. When it comes to bias, the issue isn’t unique to AI models ― humans have difficulty navigating bias as well. However, with AI, the potential outcomes of bias can have a massive impact. In AI, bias has a strong correlation with input data. For example, corrupted, unrefined, or flawed input data will impact the outcome. The important thing to grasp with bias is that it requires sensitivity, insight, and openness to navigate ethically. Humans ultimately control bias in AI ―  the users select the original input data and introduce bias to influence outcomes. For example, one major American company that receives a massive amount of job applications decided to test applying AI to its recruitment process. When it did so, the company used the résumés of current employees as input data. So, what was the outcome? The company widely shared that when using the selected demographic sampling, the results were biased against women. During the testing process, it was discovered that if the word “women” was anywhere on a résumé, that individual never got a call. The company realized the input data was part of the issue and never deployed the model for hiring managers. Sharing this information and being sensitive to the results are essential as we continue discovering the best use of this technology. Since bias is highly related to intent, the example above must not be interpreted as a malicious use of AI. Instead, it demonstrates the necessity of introspection in the use of AI. Companies can correct outcomes by factoring in bias to the model to help them achieve a more balanced result. As previously stated, AI has very quickly become an essential part of business, and it should be expected that ethical issues such as bias will occur. The keys to overcoming bias are making sure the input data is as pure as possible and being willing to investigate unethical ou t comes wi t h opennes s and transparency. In light of this, it will be necessary t o con s i de r by whom and how bias can be overcome. massive corrupted unrefined flawed recruitment malicious introspection transparency Find words which mean: • very large or heavy in size, quantity, or extent: • intended to harm or upset other people: Inference Q6 What can we deduce from the example of a major U.S. company that receives a large volume of job applications? Bias In, Bias Out 5 10 15 20 25 30 35 Fears of AI injecting bias are exaggerated.