Foreign language. Black and white. No-name cast. This is the perfect storm of descriptives to turn me off a film. But, Netflix matched it to me with a score in the 90s.
Although the online streaming behemoth is privy to all my dirty viewing secrets (full disclosure: superheroes and space wars), so am I. And I would not have opted for Roma of my own volition. But, Netflix was presuming to know something about me that not even I did. So I gave Roma a go. I was not disappointed.
Now I don’t profess to understand the inner magic behind algorithms. I suppose the in-the-know crowd is a pretty elite club of tech wizards. But there is one thing that I do get. Algorithms power machine learning and deep learning tools that make it possible to process and analyze an infinite set of data points. Something impossible for my brain, your brain, and even Einstein’s brain to fathom. And the trove of data Netflix amassed from my addictive binging habits, and then plugged into its algorithms indicated I would like Roma. Something even I was not aware of.
And for a lot of people, this is terrifying.
48 percent of the European population does not know what an algorithm is. Less than half know that algorithms are already being used in many areas of their lives.
Algorithms have gotten a bad rap: They singlehandedly triggered the downfall of democracy. They spark bad consumption behaviors. They will be responsible for the loss of hundreds of thousands of jobs. They are biased. They discriminate. Their decisions are incomprehensible and untrackable. That’s a lot of negative press for something that is basically a glorified instruction manual.
And compounding these fears are the throngs of political spin doctors, half-baked regulatory efforts, and widespread digital illiteracy. All of this is also a perfect storm. And it is perpetuating a false narrative. One that focuses a lot more attention on where algorithms have gone wrong while completely neglecting to mention where they have gone very, very right.
Algorithms Redefining Diagnostics
When the Munich Leukemia Laboratory expanded its facilities and staff, sizable investments were made on state-of-the-art Next Generation Sequencing (NGS) systems. Big machines for big data.
The science is complex, but simply put: these NGS systems make it possible to analyze the whole genome sequence (WGS) providing more data, and faster, on our genetic material. Why does this matter? The data provided by WGS holds the key to detecting pathogenic mutations, inherited disorders, and disease outbreaks. And it is this data, which machine learning applications instructed by algorithms can analyze better and faster than the human eye.
Algorithms are capable of detecting mutations and aberrations with greater accuracy than a microscope: 0.3 percent vs. 5 percent of residual leukemia cells detected following treatment.
One of the algorithms playing the biggest role in the development of NGS diagnostics is an alignment algorithm. It can align the many, and lengthy sequences of DNA and RNA determined by the NGS system. Something which would be difficult, if not impossible, for scientists and doctors to do given the sheer amount of data. Once these sequences are aligned, it is possible to draw comparisons and detect anomalies in the genetic code. And very, very quickly.
And for diseases like leukemia, where time is of the essence, speed and data can mean the difference between life and death.
“Precision medicine seeks to use genomic data to help provide the right treatment to the right patient at the right time.”
Analysis provided by alignment algorithms is advancing cancer treatment in other ways, too. The sequencing of tumors provides detailed information on the specific genetic anomalies perpetuating tumor growth. And this means oncologists can provide patients with therapies that cater specifically to their genetic mutations. This field of precision medicine has the potential to eliminate a lot of guesswork when it comes to treatment, meaning patients will no longer be forced to undergo unnecessary and complex procedures. And they will also get the best, most targeted treatment sooner, which makes all the difference when it comes to aggressive debilitating diseases.
Dressed by machines
Though not all algorithms operate with such altruistic intentions, they are consistent in their functionality. They lay out a set of rules to solve problems and perform tasks. One such task they have become infamous for happens to be something that humans are notoriously lousy at: making decisions.
That is what Stitch built a business around. It’s USP? Data science. This would hardly be unusual for a software company. But Stitch is anything but. The company is an online clothing retailer. One of thousands. But with a twist. Stitch completely assumes the shopping decision-making process.
Now there are plenty of people who might be horrified by a future where algorithms dress us. Anyone who’s read poetry written by algorithms, listened to music composed by them, and seen images generated by them has borne witness to the stuff of dystopian nightmares. Creative license should not be wholly entrusted to machines.
“Machine learning is going to change every single aspect of technology, but no machine will be able to mimic the creative ability of the human mind.” Shantanu Narayen, Adobe
But remember, algorithms can navigate data in a way that humans can not. And the more data you give them, the cleverer the learning tools they drive get. That is what Stitch has capitalized on.
The company’s styling algorithm churns through the data: customer data, product data, feedback data. Sizes, favorite colors, popular styles. But also upwards of 30 measurement attributes per item: ranging from distance between elbow to shoulder on a button down shirt to inseam length on a pair of pants.
Once the algorithms have “selected” your perfect new threads, they hand off the final decision to human stylists, also chosen by an algorithm to best suit your style. These stylists make the final call on what shows up on your doorstep.
One of the company’s goals? To eliminate the need for sizing labels and only offer each customer ‘your size’. Anyone who has have ever bought jeans sees the value in this. I, personally, would happily skip the psychological trauma and ego-shattering battle of squeezing into X-number of jeans only to go home with nothing.
Even at its most powerful, AI doesn’t replace human intelligence, it simply helps people do what they love better and faster.
But there’s another benefit. One that many people don’t consider. Let’s go back to Roma. A monochrome foreign film in a sea of superheroes and sci-fi thrillers. It can be a monumental effort to break our habits and patterns. To open our minds, our video libraries, our closets to something new. Not because we are necessarily averse to new things, or could unequivocally say we won’t like something. But because we are creatures of habit.
And though it might be creepy to some that new digital business models are employing algorithms to better predict our like and dislikes than we can, perhaps we ought to see it as an opportunity. The chance to expand our horizons, to dispense with our predilections. I, for one, am quite happy that Netflix is showing me that there is a world beyond the Marvel Universe.
Bertelsmann Stiftung, Was Europa über Algorithmen weiß und denkt Ergebnisse einer repräsentativen Bevölkerungsumfrage, February 2019
Nicholas J Schork PhD., Population Difference in Whole Human Genome Variation and Clinical Sequencing
J Pers Med, The Role of Next-Generation Sequencing in Precision Medicine: A Review of Outcomes in Oncology, September 2018
Adobe, Amplifying human creativity with artificial intelligence