Question

I worked as a mathematician and then as a quant in finance - I saw the worst of finance.

I went into data science and I was struck by what I thought was essentially a lie – namely,
that algorithms were being presented and marketed as objective fact.
A much more accurate description of an algorithm is that it’s an opinion embedded in math.
An algorithm is a very general concept - it’s something that we do actually in our heads
every day.
To build an algorithm we need only two things, essentially: a historical data-set and a definition
of success.
So I build an algorithm to cook dinner for my family.
The data that I use on a daily basis is the ingredients in my kitchen – sometimes the
time I have, the ambition I have for that dinner.
And then I assess the dinner after the fact – was it a success?
I define that because I’m the one who is building the meals – I’m in charge, I
have the power (there’s always a power element here) and I’m in charge; I get to decide
a meal is successful if my kids eat vegetables.
My kids, if they were in charge, would have defined it differently.
And it matters because we optimise over time; we optimise to success.
The succession of meals that I cook from month to month, is a very different path of meals
than if my son were in charge.
So we do that every time we build algorithms – we curate our data, we define success,
we embed our values into algorithms.
So when people tell you algorithms make things objective, you say ‘no, algorithms make
things work for the builders of algorithms.’
In general, we have a situation where algorithms are extremely powerful in our daily lives
but there is a barrier between us and the people building them, and those people are
typically coming from a kind of homogeneous group of people who have their particular
incentives - if it’s in a corporate setting, usually profit - and not usually fairness
towards the people who are subject to their algorithms
So we always have to penetrate this fortress.
We have to be able to question the algorithms themselves, especially when they are very
important to us.
We have to inject ethics into the process of building algorithms and that starts with
data scientists agreeing and signing a Hippocratic oath of modelling.
We have to stop blindly trusting algorithms to be fair - they are not inherently fair,
they are inherently picking up whatever bias we’ve given them – and start looking into
what they are actually doing. 1. What is the premise of O'Neil's argument? It is valid? to what extent?

2. What can we do to continue using this tool while trying to minimise its deleterious effects?

Answer 1

O'Neil's argument is that algorithms are shaped by creators' opinions and biases, a valid claim highlighting the lack of intrinsic objectivity in algorithmic models. To minimize biases, we must promote diversity among developers, uphold strong ethical standards, maintain transparency, and continually scrutinize and improve algorithms.

Step-by-step explanation:

The premise of O'Neil's argument is that algorithms reflect the opinions and biases of the people who create them. She asserts that they are not intrinsically objective, but rather are shaped by what we choose to measure as success and the data we use to inform them. This perspective is valid insofar as it emphasizes the subjective elements that play a significant role in the development of any algorithmic model.

To minimize the deleterious effects of algorithm biases, we must engage in a multi-faceted approach: promote diversity in the teams developing algorithms to ensure a range of perspectives and experiences; instill a practice of rigorous ethics in data science, similar to a Hippocratic oath for medicine; maintain transparency in algorithmic design and decision-making processes; and actively search for and correct biases within these systems. Additionally, it is important to have open dialogues about the impact and governance of these technologies in society to better understand how they can be used responsibly.

Considering the increasing reliance on algorithms in everyday life, from search engines like Go_ogle to social media platforms like Facebo_ok and beyond, understanding the underlying biases and fostering ethical practices becomes especially critical. Algorithms are tools for interpreting data and making decisions, and they must be scrutinized and improved continually.