What I Learned From Completing the MIT xPro Data Science Course: Part 2
The professional insights, about data science and machine learning
Part 2: What have I learned about data science and machine learning?
About the power of new toys
I first learned maths on the abacus in primary school in the 80s in China, no kidding. Then I learned and worked in Excel, where I thought, how on earth did people ever live without it? Now, how Excel feels like to Python is just like how the abacus was to the Excel. Wow, isn’t that something.
About “clustering”
It is hard to look at Facebook, Netflix, and Google the same way again. Why do we see the feeds we do? I am now more conscious of the recommendation algorithms working behind every click. Our world is so filtered. We are “clustered” like the machines program us to be. No wonder we are so divided. When we use social media and search engines without awareness, they create powerful confirmation biases.
Isn’t that ironic? Technology has connected continents, and yet as a race, humans are perhaps further apart than ever. To experience humanity and the surprises of life, there perhaps has never been a greater need for more discerning human judgement, and to challenge ourselves to walk away from the screen.
About the stakes in data and AI
One of the greatest machine learning problems was solved via the global data science competition platform, Kaggle. And it was to improve the accuracy of the recommendation algorithm at Netflix by, a insignificant sounding, 10%. It took 3 years for the winning team to finally emerge in 2009. And the $1m prize awarded was said to be money well spent. Such are the things at stake in AI, 10% for $1m. And such is the magic of competition.
About ANN
Artificial neural networks (ANN) have the major vulnerability, that when the layers are deep, and when we don’t understand why it works, tiny disturbances can cause wholesale errors. So General Artificial Intelligence? The ultimate expression of an ANN. What does this mean about that? There is perhaps no silver bullet.
About the seed 42
A popular seed for setting the random state of many algorithms is 42, because it is the Answer to the Ultimate Question of Life, the Universe, and Everything. Amen to 42.
About the boundaries of software
AI is a data and software technology which started post WW2 in the 1950s. It however experienced an “AI Winter” in the 1970s, where its development was stopped by hardware constraints, due to memory and processing power not being good enough to carry out the AI projects.
Fascinating relationship between hardware and software. The boundaries of software are set by the hardware. Perhaps this was why Steve Jobs took Alan Kay’s idea seriously, that, “People who are really serious about software should make their own hardware." And today, we have the Apple devices in our hands to prove it.
About social credit
There was an intriguing episode of the sci-fi TV series Black Mirror, called “Nosedive”. It was about a possible future social order decided by our digital social credit or popularity. The story makes sense to me now. It is where the algorithms are leading us.
How far will we go to get the currency of an influencer? Interesting ethical question. But as some may say, “there are no new problems”. New currencies are still currencies. So this could be an age old question, about how to treat currencies as means vs ends.
About the “End of Equality”
It took a great deal of work for me to learn this course, and it makes me think of people who do not have the prerequisite knowledge to gain the skills I did. I am reminded of a recommended book on AI, 2062: The World That AI Made, by Toby Walsh. It predicts a trend to the “End of Equality”, where AI would concentrate “wealth and power in the hands of the technological elite”, unless there is “corrective action”.
I am not a maths genius, I never did go from the actuarial associate level to the fellowship. But I am fortunate to have had more mathematical training than the average person. Knowing how much it has taken me to re-train myself, I can see why Toby Walsh arrived at his view.
There could perhaps never be a clear solution. The future is always an unchartered territory. In Ray Dalio’s words, “AI will lead to … an exciting and perilous new world. … And as always, … we are much better off preparing to deal with it than wishing it weren’t true.”
About “the Rationalists” who work on AI
Reflecting also on how much of a laser focus it took, I am reminded of another book on AI, called The AI Does Not Hate You, by Tom Chivers. It describes a global community of people most involved in AI work, called “the Rationalists”.
They are “nerds”, “deeply interested in how things work”, “disproportionately male, and disproprotionately on the autistic spectrum or near to it”. They “lack social skills”, experience social difficulties, and are often misunderstood.
I can somehow see them now, and see the kind of focus and maybe isolation it takes to be with machines and work on AI. I can see now the humanity in their difficulties too. It is humbling.
About being ourselves
I have realised through this, it is important for me to keep my dancing and yoga, because there is so much humanity to be discovered there, and also they are more “me”. I used to feel conflicted between being “me” and being “different”, in that “How am I ever gonna compete with them?”
But now, in recoginising, allowing, and not being fearful to live my own differences, it helps me reconcile with and appreciate those who are different from me, those who work single mindedly on AI. Because perhaps they are fulfilled by their narrow focus in the same way I am by my diverse exposure. I did use to make right and wrong out of this.
About competition vs team, and the bigger picture
Technology will perhaps both fulfil more human potetial and individualities on the one hand, and magnify more human differences on the other. But having this experinece makes me see the humanity that remains timeless.
We still share the same desire to fulfil ourselves, to be respected and accepted as a human being, and the same difficulties to live with the human condition. There will always be a time to compete to become better in ourselves, and always a time too to team with a diverse people, to be more than ourselves.
There is a peace, equanimity, humility, and power in recoginising our humanity.
The series
This is the second of a 3-part series. For the summary post, and the other two parts, please find their links below.
- Summary post,
- Part 1: The personal insights, on taking up a challenge and a new learning project; and
- Part 3: The modern way of learning, on virtual learning platforms, comparing MIT xPro vs popular alternatives like Udemy.
