Case Study
- What are two reasonable hypotheses that would explain why the machine learning-driven ad targeting system presented ads for high-paying jobs to men at a much higher rate than it displayed them to women? Be specific and use concepts from course readings to elaborate on your hypotheses.
One of the hypotheses that can be used to explain the reason why Google was showing highly paid CEO positions to men and less to women is that the company attempts to have the best answers for each person as they search for information online through the programming of its algorithms. This is one of the strategies that Google has implemented to try and maximize the experience of the users online. By doing this, Google wanted to ensure that searchers get the most relevant information featuring the answers that they are expecting. The company does this by collecting information regarding the interest of the users and well-known information about the persons, such as their gender. According to the Carnegie Mellon research, it was evident that accounts that had gender as female did not get adverts for high paying CEO jobs after searching. It is extremely hard for a searcher to get information that is not biased from google since it is difficult to circumvent the personalized online searches. There is little that searchers can do to avoid the personalized search, which seems to be embedded in the algorithms of the company. The algorithms of the company are designed in such a way that even users who log out will still get customized results for their search. This, therefore, means that they will get shaped results from goggle. The changes that were made by Google in 2009 made the personalized search default for everybody, including those who had logged out. This is one of the probable sources of the problem that exists with adverts. Therefore, this hypothesis postulates that the error of biased results from google search can be explained by the anomalies in the company’s algorithms.
Another hypothesis is that the customers who order for the adverts chose the people to view the ads. This might be the case because the employers are likely to have a target clientele for the adverts. Therefore, this places a duty on google to ensure that the ads are only viewed by the candidates who are requested by the customers. Just like the recommendation that is made after the ad fisher study, that the company will only operate within the framework of its policies, therefore google will only follow its guidelines. If the policies of the company state that the adverts will be customized as per the request of the customers, this should happen. Most employers target candidates for the high paying jobs, in the labor markets where candidates are likely to possess not only the relevant qualifications but, most importantly, have the experience that is needed in such positions. Furthermore, most of the customers will target male applicants, especially if it is an oversee assignment because they are more mobile than females who may need to take care of the family. These hindrances can be used to justify the claim by google that the adverts are to be shown to people whom the advertisers requested. It is wrong for the advertisers to advertise, and the ad fails to attract the people they target. Therefore, discrimination in the adverts may be done intentionally, by google as it tries to meet the requests of the advertisers.
- Given each hypothesis, describe concrete steps, Google could take to minimize the likelihood of similar problems emerging in the future. Describe two recommendations for each hypothesis, for a total of four concrete recommendations.
The solution to the issue of manipulation of the algorithms is that the company needs to work on the algorithms such that they do not discriminate against a certain group of people either by race or gender. This calls for radical programming of the google search so that it can eliminate the errors. The algorithms used by Google will only work as human programmers have instructed them. Therefore, the company needs to program the algorithms such that this bias is eliminated. This can be done by determining the type of algorithms used, how they are set up, the metric and the parameters to be used, and determining and testing what the algorithms are designed to do.
To further eliminate the filtering bumble, the company needs to have an option for the users to decide whether they would like to use the personalized search method or not. This will be prudent since people will be using the filtering bumble at their volition. It will also ensure that customers do not complain about getting biased results.
To solve the problem of customer requests, the company should have a policy that allows for all the adverts to be seen by all people irrespective of their gender. This is because modern women are educated and equally competent. This will compel the advertisers to filter the candidates from their end as opposed to hiding the ads to women. The policies of the company should be supreme to the request of the customers. Customers should work within the framework that is set by the company such that people who visit google will not be offended by the working of the algorithms.
Another solution to the problem of advertising is for the company to define its ethical principles so that it will be clear what is allowable and what is not. It is unethical for the company to discriminate against people because of gender or color. The ethical code should be included in the algorithms of the company, such that they will not discriminate against people on the bases of gender or color. The company should assess the workplace to determine what ethical issues should be resolved in the company and the system so that it can develop solutions both to the human errors that may contribute to this scenario and the problems with the algorithms.
- What is a fair way to target job advertisements like this? Be specific.
The fair way of advertising for jobs, in this case, is to ensure that the adverts get to all people so that those who are qualified will apply for the positions. Contemporary women are educated, and they are able to perform duties that were previously thought to be for men. This will guarantee that the algorithms do not discriminate against women or people of other races against applying for the position. This will create the perceived fairness in the advertisement process. It is unethical that online ads only target men to take the top spots in companies. To achieve this end, the company needs to ensure the algorithms that it uses are tested extensively for the contextual validity to ensure there is a very high legal and ethical impact. The company should allow for the algorithms to be verified by third parties so that it eliminates any form of biasness so that the adverts will reach all the people equally.