Algorithms Rule Us All: Evaluating The Biases And Risks Of Machine Learning

Evaluation Of The Documentary

The documentary to be reviewed is “Algorithms Rule Us All”. The public broadcaster behind this documentary is Vrijzinnig Protestantse Radio Omroep (VPRO). This is a Dutch public broadcaster that forms a part of the Dutch public broadcasting system. VPRO was founded in the year 1926. Algorithms Rule Us All was broadcasted in the year 2018 and is directed by Martijn Kieft (Kieft 2022). The documentary is addressed towards data scientists and programmers seeking an answer to the question of what their mindset is about Big Data society. The content is centered around how human lives and societies are going to surrender to the urge of machines that again are driven by profit-driven firms. There is no doubt in the fact that an algorithm’s ability to carry out the processing of volumes of data points may be helpful when putting in efforts to organize waste management. But instances such as Google’s algorithm advertising higher-paid jobs to men is a matter of concern. The question of biased programs is a big debate and it certainly needs to be understood as to what extent algorithms should be allowed interference in human lives. Thus, in this reaction paper, the documentary will be evaluated and some insights drawn from the content shared with the readers.

With The Rapid Advancements Happening In Science And Technology It Can Be Said That Computer Systems Are No More To Just Help Humans But Rather These As Well are taking important decisions of human lives. But the concern is whether these algorithms should be given all of the control. Do these algorithms always make better decisions? In the documentary, it is said well that people at first need to understand that it is not the algorithm that is controlling human lives but rather it is the company that has devised or developed the algorithm (Kieft 2022). The company can make undue use of the data that is collected through the algorithm developed by it. This can be well supported by some reports that say companies are making quick moves to apply machine learning to business decision-making (Controlling machine-learning algorithms and their biases 2022). The speed at which these algorithms are being developed can be attributed to the attractiveness of the technology, but here the lack of experience does create major risks. When speaking about risks, it is “Algorithm bias” that has to be mentioned (Controlling machine-learning algorithms and their biases 2022).

Big Data—Solution or Curse?

It had become a notion that big data and AI will save human lives, assuming that data will be fed and results obtained will reveal what should be done. But this certainly is not something that will save human lives. This is so because data is not neutral and has stereotypes or is influenced by concepts that lead to what is called Algorithm Bias (Kieft 2022). It has to be accepted that there is biasness in the process of data collection and data interpretation. The central fact is that the algorithms are developed by humans and these humans have some ideologies or concepts that overpower their thought processes. Had it not been so, then Google algorithm would not have suggested high-paid jobs to men only. The biased algorithms if thought about intellectually will lead us to the eye-opening fact that inequality has deep-rooted itself in present-day society (?Algorithmic Bias: Why And How Do Computers Make Unfair Decisions? 2022). This needs to be supported as when training an algorithm to perform a specific task, humans feed it with the data that reflects the happenings in the society at the exact moment or in the past (?Algorithmic Bias: Why And How Do Computers Make Unfair Decisions? 2022). To understand it better, the example of an algorithm that screens job applications can be considered. Here there are high chances of the criteria for what kind of CV is a good match is based on who have got those jobs at the moment. If the algorithm detects that almost all serving the job role are white men, it can tend to believe that the ideal candidate for the post is a white man.

There are other factors been mentioned in the documentary as well such as missing data for example much of the crime data is not available and arrest data serves as the proxy (Kieft 2022). Big names in the market mean some well-known companies are taking the advantage of these algorithms to make people addicted to some things and also make use of the same for manipulation. They are emphasizing negative emotions like that are more powerful than positive ones and thus in all ways, it can be said that in this age of algorithm human lives are in danger (Kieft 2022). Being guided by the algorithms will mean decisions being taken by these smart machines and humans falling prey to it. Humankind is taking much pride in the fact that they have come up with inventions such as robots and harnessed technologies such as automation. But the fact is they are missing the link of being a slave to these algorithms. It has to be accepted that algorithm bias comes from the choices that the developers make when creating the algorithms rather than from an extremely prejudiced motive. Here yet another term of Algorithm Transparency needs to be mentioned. Much has been said about the biasness of algorithms but then the roots are certainly the human thoughts. Framing in this way will be better than as compared to humans, algorithms developed can be more honest simply because here the selections are dependent on math and data instead of being dependent on fuzzy human logic and associated emotions (Algorithm Bias Explained 2022). With proper algorithm transparency and practices of record keeping, a regulator can figure out the various factors that have been considered by the algorithm to make a certain decision. Contrary to this, when humans are biased, they can straight away deny the fact and all is hidden. But in the case of algorithms that cannot happen. Biased or not it will all be reflected in the code or the training data.

Big Data—Solution or Curse? for companies

If the documentary is taken as the reference point then it can be said that Big Data is turning out to be a curse. The challenges, in this case, are insufficient understanding and acceptance of big data. It is a fact that employees do not understand the value of big data and the next big issue is being confused with the various technologies of big data. Adopting Big data again requires much investment to be done and thereafter it is also complex to manage the data quality. Lastly, the issue of security is already there and is one of the most serious concerns with big data. Data scandals such as Cambridge Analytica are certainly evidence of the fact that Big Data can turn out to be a danger in the future (Cambridge Analytica and Facebook: The Scandal and the Fallout So Far (Published 2018) 2022). This was the largest know leak of data in Facebook history. Some employees of the firm sell psychological profiles of American voters to political campaigns, acquiring the private Facebook data of tens of millions of users. This is an alarm and people should understand how the advancing technologies that are being developed to help humankind are making them slaves.

Big data and automation have to be taken to be as harmful to society and the organizations using these because of the shift of control being brought in (Ewert, Brodowicz, and Pospieszny 2018). From the documentary, it is clear that gone are the days when a computer was just there to help humans with their work mostly repetitive tasks (Kieft 2022). The situation today is decisions are being taken based on algorithms and human thinking capability being constricted. Machines that were developed to follow commands have all of the control now (Nadikattu 2020). Big data and automation can thus be dangerous to organizations if their power is not harnessed well. With the advantages of these technologies, it has to be ensured that disadvantages be considered as well. Proper steps need to be taken to address the issues in place before going on to step ahead. Just for the sake of using Big Data technologies, companies should not rush with the decision to incorporate it into their operations (Al-Sai, Abdullah, and Husin 2020). With the increasing leak of data, even tech giants such as Facebook are losing hold on their market. The negative image is being developed and thus it is high time to realize the way the power of Big Data and automation be harnessed. There is no doubt that technologies as such will help businesses to venture out in new markets but then without taking note of the dangers associated with these technologies, many turbulences have to be faced for sure.

Privacy issues are one of the major concerns in this age of Big Data and automation. For instance, when using Facebook much of the user data is collected. The surfing history of the users reaches the tech giant and as per it, suggestions are provided. As cited in the documentary the algorithm of Facebook is well developed to determine which political advertisements an individual sees. Starting from the application procedure to a dismissal request, all of these are controlled by an algorithm. Users’ location, preferences, and many more of the details are being noted down by these algorithms thus breaching their privacy. As per the reports, most of the customers that is almost 74% of them think that they have little or no control over how their data is being used. Amongst them, only 40% think that the industry can bring in self-regulations when it is the question of privacy issues (Most consumers do not trust big tech with their privacy 2022). The primary privacy concerns revolve around losing civil liberties and having data related to identity stolen or sold.

There are again ethical issues as well with the use of these modern technologies or algorithms. Though much of the repetitive part of tasks can be done easily with algorithms it cannot be denied that there are some serious ethical concerns as well. As already mentioned algorithms work on train data and lack human empathy. These algorithms provide results based on data that again is not neutral. Thus, here as well the examples from the documentary have to be agreed with. For instance, an American prisoner had to wait for long as compared to others as the algorithm establishing a risk score gave him a strangely high outcome (Kieft 2022). Again, unlike a judge’s verdict, the decision made by an algorithm cannot be challenged. Someone who deserves to be free has to sit back and this is highly unethical. Here, the ethical issues of lack of transparency of the algorithmic process, unfair results, and unintended costs are evident. 

Conclusion

Thus, it needs to be agreed that “Algorithms Rule Us All”. It is high time to understand that companies developing these algorithms are the actual ones to breach people’s privacy. Awareness is something that is much needed in this context. Algorithms being developed are required to be transparent so that if needed they can be evaluated considering some proper aspects. Most importantly, there have to be proactive steps taken in the direction of framing laws and rules which evaluate whether an algorithm should be released in the market or if it should not be. The main points of this discussion are that Algorithm Bias is the result of the societal inequalities in place. The developer’s mindset can be related to the biasness that is brought about with the algorithm. The examples of men being shown high paid jobs and white men being selected as eligible candidates for a post suggest that the algorithm just works on train data and history and the liability is that of the person who has developed it.

References

1. Controlling machine-learning algorithms and their biases. [online] Available at:
2. <https://www.mckinsey.com/business-functions/risk-and-resilience/our-insights/controlling-machine-learning-algorithms-and-their-biases> [Accessed 27 March 2022].

Al-Sai, Z.A., Abdullah, R. and Husin, M.H., 2020. Critical success factors for big data: a systematic literature review. IEEE Access, 8, pp.118940-118956.

Ewert, M., Brodowicz, D.P. and Pospieszny, P., 2018. Artificial Intelligence and machines: A curse or blessing for corporate real estate?. Corporate Real Estate Journal, 7(4), pp.337-351.

Greenlining.org. 2022. Algorithm Bias Explained. [online] Available at:

<https://greenlining.org/wp-content/uploads/2021/04/Greenlining-Institute-Algorithmic-Bias-Explained-Report-Feb-2021.pdf> [Accessed 27 March 2022].

Kieft, M., 2022. Algorithms rule us all – VPRO documentary – 2018. Available at: <https://www.youtube.com/watch?v=NFF_wj5jmiQ> [Accessed 27 March 2022].

Liberties.eu. 2022. ?Algorithmic Bias: Why And How Do Computers Make Unfair Decisions?. [online] Available at:

<https://www.liberties.eu/en/stories/algorithmic-bias-17052021/43528> [Accessed 27 March 2022].

Nadikattu, R.R., 2020. Research on data science, data analytics and big data. INTERNATIONAL JOURNAL OF ENGINEERING, SCIENCE AND, 9(5), pp.99-105.

Nytimes.com. 2022. Cambridge Analytica and Facebook: The Scandal and the Fallout So Far (Published 2018). [online] Available at: <https://www.nytimes.com/2018/04/04/us/politics/cambridge-analytica-scandal-fallout.html> [Accessed 27 March 2022].

TechRepublic. 2022. Most consumers do not trust big tech with their privacy. [online] Available at:

<https://www.techrepublic.com/article/most-consumers-do-not-trust-big-tech-with-their-privacy/> [Accessed 27 March 2022].