Scribd
Upload
74 views2 pages

Synthèse Mathilde PDF

The document discusses the controversy around algorithms and where the appropriate limits of human reliance on them should lie. It analyzes various sources that both praise algorithms for their effectiveness and impartiality when based on large datasets, but also criticize their overuse and lack of transparency. While algorithms work well for recommendations and predictions, they have limitations in understanding causality and explaining decisions. The ethics behind how and why algorithms are used is important to consider, as companies prioritize profits over user impacts. Overall, algorithms are best suited for predictive tasks with quality data, but humans must understand their capabilities and ensure proper oversight to avoid issues like the flawed university admissions algorithm.

Uploaded by

Mathilde Urbain
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
74 views2 pages

Synthèse Mathilde PDF

The document discusses the controversy around algorithms and where the appropriate limits of human reliance on them should lie. It analyzes various sources that both praise algorithms for their effectiveness and impartiality when based on large datasets, but also criticize their overuse and lack of transparency. While algorithms work well for recommendations and predictions, they have limitations in understanding causality and explaining decisions. The ethics behind how and why algorithms are used is important to consider, as companies prioritize profits over user impacts. Overall, algorithms are best suited for predictive tasks with quality data, but humans must understand their capabilities and ensure proper oversight to avoid issues like the flawed university admissions algorithm.

Uploaded by

Mathilde Urbain
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Urbain Mathilde PSIE

The controversy about algorithms

People often criticize pricing algorithms as they don’t care for them and their situation as
individuals. This controversy about algorithms, amplified by the Offqual’s algorithm fiasco, is discussed
in the set of documents, which includes two articles from the MIT Technology Review and The Guardian,
respectively written by Karen Hao in last August and by John Naughton, a song from the punk album
Sisteray Said called “Algorithm Prison”, and an opinion piece written by Rob Francis for the Medium
website. Where should lie the limit of human reliance on algorithms? First, we’ll explain how and why
humans rely on algorithms, sometimes not even realizing it, then we’ll study the limitations of algorithms,
so as to enlarge on the question of when it does become inappropriate to rely on them.

On the one hand, Naughton and Francis make clear in their articles that when it comes to making
predictions based on past behaviors, the use of algorithm has become pathological over the years and have
proved effectiveness and impartiality, since they are regularly reviewed, tested, and rely on years of data
gathering. The Ofqual’s A-Level results algorithm fiasco, discussed by Hao and mentioned by Francis,
goes to show it. On the other hand, they are ways in which people not only aren’t aware they are using
algorithms but also they don’t realize it. Indeed, in his song “Algorithm Prison”, Systeray highlights the
loop London habitants are trapped in: they’re blindly repeating the same humdrum routine, anesthetized
by an algorithm that defines their life.

However, although they work well in the offline world, algorithms have limitations. As Naughton
explains, machine-learning can produce algorithms unsuitable if the need is to understand relationships or
causality. Moreover, their inability to explain their decisions to human users can make it difficult to use
algorithms. For instance, when it comes to creating cookies or recommendations online, using algorithms
works well as the user’s feed constantly adapts itself to his or her preferences and interests. On the
contrary, as Hao and Francis explain, the Ofqual algorithm, which had been created to fairly assess
students for university admissions, went wrong because of two factors: the lack of quality data, tests, and
reviews, but also the ethics of how it had been programmed.

In fact, much more their effectiveness, it’s the ethic behind the use of algorithms we should
genuinely question. Indeed, as sung in “Algorithm Prison” and explained in Naughton’s article, giant tech
and advertising companies put ML everywhere online to make money, but they don’t care about the user’s
safety or sanity: their purpose is to target the user in order to make them spend time on their site. At last
but not least, Hao and Naughton point out the common belief any algorithm will fix a broken system, but
in the end, it suffers from it, like when the secretary of state himself prioritized the need to standardize
grades instead of helping students transition into university, while it’s the ultimate purpose of entrance-
exams.

To conclude, when limited to a strictly predictive job and in possession of loads of quality data,
using an algorithm assures effectiveness and impartiality. The line not to cross so as to keep it that way
only concerns humans and both their utilization of and their expectations from the algorithm.

(550 words)
Urbain Mathilde PSIE

You might also like