Man vs. Machine: Are Algorithms Or Human Curation the Answer?

18 Dec 2015


Whether or not you’re aware of it, countless aspects of your life are controlled by algorithms, which play a huge role in determining everything from your next favorite movie and can’t-get-it-out-of-your-head song to your future purchases and even partners.

We’ve already detailed the power of algorithms in a previous blog, Algorithms: Powering Search Engines, Finding Your Soulmate and (Maybe) Stealing Your Job. As a reminder, algorithms are utilized in everything from e-commerce to online dating and music streaming, and an algorithm can be defined as “a process or set of rules to be followed in calculations or other problem-solving operations, especially by a computer.” (Yes, even a computer thinks your love life is a problem that needs solving.)

However, despite the obvious benefits of implementing algorithms—efficiency, scalability, etc.—there is a definite question about cases in which algorithms aren’t appropriate and the situations when you really need a human touch: When it comes to creating a personalized experience and providing users with content that’s relevant, can computers or people do it better? Human curation is going through a resurgence in popularity, but there is still always the chance that algorithms could take over—especially given recent AI open-sourcing announcements from Google and Facebook and the news that Elon Musk and more would be investing a cool $1 billion in artificial intelligence.

It turns out that both humans and machines have their strengths, and what you’re looking for and how you’re looking for it might determine whether it’s the former or the latter that’s better suited for your needs.

Who’s Your DJ?

This debate about which is better, algorithms or human curation, is nicely illustrated by the contradictory approaches of Apple Music and Spotify.

Spotify’s Discover Weekly—a favorite feature of music lovers hungry to discover their next favorite artist—is reliant on algorithms, which build a taste profile for each user based on what they listen to, the songs they add to playlists, and the music they choose to interact with. Spotify then takes this taste profile to find similar users and see which songs or artists they have interacted with or added to playlists. Spotify cross-references the taste profile with genres that they believe user is interested in exploring, and the result is a playlist that recommends new songs to the user each week. As for the effectiveness? Tech entrepreneur and web pioneer Anil Dash told The Verge that the machines manning this experience were “as good as DJs — at scale.”

Apple, meanwhile, rejects this method, insisting that computers can’t really capture the emotional aspects of music and are incapable of creating playlists that really flow, hitting and building to emotional beats. To combat this, Apple Music is curated by several well-known DJs and artists, like Zane Lowe and Drake, who can opt to play specifically themed music and do what machines have trouble doing: present songs and artists before they become popular. (Oh, Apple…so hipster.)

Case by Case

Besides Apple Music and Spotify, there are plenty of other situations that advocate for algorithms over humans or vice versa.

Editors at NYMag, for example, conducted an experiment pitting third-party algorithms against editors to determine which were better suited to select and gather links to hook users and cause them to read the most articles. It turned out that the editors took the curation cake in this situation and the links they chose “significantly outperformed” those generated by the algorithm. The outlet is now using algorithms to assist in choosing articles to recommend, but the bulk of the responsibility is assigned to editors to choose which content is promoted on websites. As Michael Silberman, the general manager of digital media at New York magazine told Mashable, “Humans are pretty good at figuring out what other humans are going to be interested in clicking.”

When it comes to hiring, howeverat least for low-skill, service sector jobsalgorithms have proven to be better. One study examined 15 companies and over 300,000 hires, and after the research, they arrived at the following conclusion:

People want to believe they have good instincts, but when it comes to hiring, they can’t best a computer. Hiring managers select worse job candidates than the ones recommended by an algorithm.

So basically, just because someone went to your alma mater and shares your affinity for seamlessly integrating Seinfeld references into everyday conversation, it’s probably in your best interest to ignore your gut and listen when the computer says the other guy is a better fit.

With Their Powers Combined…

So would youor companies, ratherbe better off implementing an algorithm or enlisting the help of fellow humans? The answer is that both of them have their strengths and their weaknesses.

Benedict Evans of Andreessen Horowitz, a venture capital firm in Silicon Valley that invests in technology companies, discusses where the split between curation and algorithm might occur. For example, Google, Facebook, Amazon etc. do a great job delivering to users what they already know they want, and combing through a database quickly and efficiently to find specific information is a task that absolutely requires algorithms. However, they are less effective at determining what users might want, whereas humans are much better at this.

That being said, in the end, your best option might be combining the two, and utilizing both the scalability of algorithms and the inherent understanding of humans to ensure the best possible outcome. After all, while can present you with plenty of date options, your friends are probably better judges of character…or at least slightly more qualified than a computer to determine if someone seems like a murderer.

Both of which, you know, are pretty valuable services.